Document template for NOBEL by zk8FVK38

VIEWS: 8 PAGES: 144

									                                                         First Guideline for Technologies
                         IST IP NOBEL "Next generation
                         Optical network for Broadband               ee787c6d-c81c-433e-a3ea-
                              European Leadership"                          3ab4c127c833.doc




Deliverable 20
Work Package 7
“First Guidelines for Technologies Enabling
broadband optical networks”

Status and Version:          PIR-a-Version 1.0
Date of issue:               30.03.2005
Distribution:                Project Internal
Author(s):                   Name                           Partner
        Leading Editor      Gottfried Lehmann              Siemens
                            Marc-Steffen Wrage             Siemens
                            Wolfgang Schairer              Siemens
                            Stefano Santoni                Pirelli
                            Dario Setti                    Pirelli
                            Lamia Meflah                   UCL
                            Bernd Bollenz                  Lucent GmbH
                            Marco Romagnoli                Pirelli
                            Eugenio Iannone                Pirelli
                            Andrea Romano                  Pirelli
                            Arianna Paoletti               Pirelli


Checked by:




                                  Page 1 of 144
                                                    First Guideline for Technologies
                    IST IP NOBEL "Next generation
                    Optical network for Broadband            ee787c6d-c81c-433e-a3ea-
                         European Leadership"                       3ab4c127c833.doc




Executive summary




                            Page 2 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                              3ab4c127c833.doc




Table of Contents
1   Introduction                                                                                     6
    1.1        Purpose and Scope                                                                     6
    1.2        Cooperation with other workpackages                                                   6
    1.3        Reference Material                                                                    6
       1.3.1   Reference Documents                                                                   6
       1.3.2   Acronyms                                                                              6
       1.3.3   Definitions                                                                           8
    1.4        Document History                                                                      8
2   Document overview                                                                                8
3   Socio-economic and technical environment                                                         9
    3.1        Technological progress                                                                9
    3.2        Customer demands and market trends                                                    9
    3.3        Considered model networks                                                             9
    3.4        Technologies to enable future dynamic optical networks                                9
4   Current component developments                                                                  10
    4.1        Technologies to increase system performance                                          10
    4.2        Technologies to enable new functionalities                                           10
    4.3        Technologies to lower CAPEX or OPEX                                                  10
    4.4        Approaches to deliver products of higher integration                                 11
    4.5        Technologies and components considered within deliverable D19                        11
5   Transmitter/receiver and alternative modulation formats (DQPSK etc.)                            12
6   Tunable Laser                                                                                   13
    Tunable Lasers Requirements                                                                     13
    Tunable Lasers overview                                                                         16
    DFB (Distributed Feedback) Tunable Lasers                                                       18
    DBR Tunable Lasers                                                                              19
    Laser Array Tunable Lasers                                                                      19
      SANTUR (DFB array)                                                                            20
      AGILENT (DBR array)                                                                           25
      NEC (DFB array)                                                                               27
      NTT (DBR array)                                                                               29
      QDI DFB array                                                                                 31
    Multi-Section DBR Tunable Lasers                                                                32
      AGILITY (SG-DBR)                                                                              33
      BOOKHAM (DS-DBR)                                                                              44
      NTT (SSG-DBR)                                                                                 48



                                        Page 3 of 144
                                                                First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband             ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                        3ab4c127c833.doc




      ADC/Altitune (GCSR)                                                                     50
      SYNTUNE (MG-Y modulated grating Y-branch laser)                                         51
      INTUNE (fast switching laser)                                                           54
    External Cavity Tunable Lasers                                                            55
      IOLON (ECL)                                                                             56
      INTEL (ECL)                                                                             60
      PRINCETON OPTRONICS (ECL)                                                               67
      ALCATEL/(AVANEX) (ECL)                                                                  68
      PIRELLI (ECL)                                                                           69
      VCSEL Tunable Lasers                                                                    69
    Tunable Lasers based on other technologies                                                70
    Tunable Lasers comparison                                                                 70
References                                                                                    72
7   ROADM – Reconfigurable Optical Add-Drop Multiplexers                                      72
    7.1        ROADM main characteristics                                                     73
       7.1.1   Connectivity                                                                   73
       7.1.2   Optical Characteristics                                                        75
       7.1.3   Architectural and Operational Characteristics                                  77
    7.2        ROADM structure                                                                80
    7.3      Mux/demux + 2x2 switch                                                           82
       7.3.1 Overview of existing products                                                    84
       7.3.2 Existing technologies and further developments on basic building blocks          89
    7.4      Mux/demux + NxN (NxM) switch                                                     95
       7.4.1 Overview of existing products                                                    96
       7.4.2 Existing technologies and further developments on basic building blocks          97
    7.5      Wavelength Blocker                                                              102
       7.5.1 Overview of existing products                                                   103
       7.5.2 Existing technologies and further developments on basic building blocks         106
    7.6      Serial Tunable OADM                                                             110
       7.6.1 Overview of existing products                                                   111
       7.6.2 Existing technologies and further developments on basic building blocks         114
    7.7        Wavelength Selective Switch                                                   114
    7.8      Other structures/technologies for ROADM applications                            125
       7.8.1 FWM in SOA                                                                      125
       7.8.2 Photonic Crystals                                                               126
8   Raman amplification                                                                      128
    8.1        Raman Gain and Raman Amplification                                            128
    8.2        Noise Components and Non-linear effects                                       128
    8.3        LRA in Transmission Systems and comparison with EDFA                          128
    8.4        LRA Products                                                                  128
9   Behavior of amps and amp chains                                                          129




                                        Page 4 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




10 Adaptive filters PMD, fast     -filters                                                       130
11 Tunable filters                                                                               131
  11.1      MEMS-based tunable filters                                                           131
     11.1.1 AXSUN Fabry-Perot MEMS tunable filter                                                131
     11.1.2 IOLON Diffraction Grating MEMS tunable filters                                       131
  11.2      Compliant MEMS-based tunable filters                                                 131
     11.2.1 SOLUS MICRO C-MEMS tunable filter                                                    131
  11.3        Tunable Thin Film Filters                                                          131
     11.3.1   AEGIS TFF tunable filter                                                           131
     11.3.2   MICRONOPTICS TFF tunable filter                                                    131
     11.3.3   SANTEC TFF tunable filter                                                          131
  11.4        Liquid Crystal tunable filters                                                     131
     11.4.1   DIGILENS LC tunable filter                                                         131
     11.4.2   NANOOPTO LC tunable filter                                                         131
     11.4.3   LC as cladding of SOI waveguides                                                   131
     11.4.4   Liquid Crystal in combination with Photonic Crystals and Photonic Crystal
     Fibers   131
  11.5        Tunable Fiber Bragg Grating (FBG)                                                  131
     11.5.1   AOS tunable FBG                                                                    131
     11.5.2   FBG Strain tuning                                                                  131
     11.5.3   ALNAIR tunable FBG                                                                 132
     11.5.4   PLC Bragg Grating                                                                  132
  11.6        Volume Bragg Grating (VBG) tunable filters                                         132
  11.7      PLC with micro-resonators tunable filters                                            132
     11.7.1 Tunable Drop filter with micro-rings in HIC materials                                132
     11.7.2 Polymer micro-rings                                                                  132
  11.8      Acousto-optic tunable filters (AOTF)                                                 132
     11.8.1 Etched cladding AOTF                                                                 132
     11.8.2 Acousto-optic polarimeter                                                            132
  11.9        Other tunable filter technologies and commercial devices                           132
     11.9.1   JDSU tunable filter                                                                132
     11.9.2   OPTUNE tunable filter                                                              132
     11.9.3   DICON tunable filter                                                               132
     11.9.4   All-Fiber tunable devices                                                          132
     11.9.5   Sampled Grating in InGaAsP/InP deep-ridge waveguide                                132
12 Compensators/FEC                                                                              133
13 Wavelength–Conversion                                                                         134
14 Regeneration: 2R/3R                                                                           135
15 Optical switching                                                                             136
16 OPM and signal quality analysis                                                               137
17 Summary                                                                                       138
18 References                                                                                    138




                                        Page 5 of 144
                                                            First Guideline for Technologies
                            IST IP NOBEL "Next generation
                            Optical network for Broadband            ee787c6d-c81c-433e-a3ea-
                                 European Leadership"                       3ab4c127c833.doc




 1       Introduction

 1.1     Purpose and Scope


 1.2     Cooperation with other workpackages




 1.3     Reference Material

 1.3.1     Reference Documents




 1.3.2     Acronyms


ADC        Analogue-to-Digital Conversion
ASE        Amplified Spontaneous Emission
B2B        Back-to-back
BER        Bit Error Rate
CD         Chromatic Dispersion
CSRZ       Carrier Suppressed Return to Zero
DB         Duo-Binary transmission
DCF        Dispersion Compensating Fibre
DEMUX      Demultiplexer
DFB        Distributed Feedback Laser
DFE        Decision Feedback Equaliser
DGD        Differential Group Delay
DOP        Degree Of Polarisation
DQPSK      Differential Quadrature Phase Shift Keying
DSP        Digital Signal Processing




                                       Page 6 of 144
                                                            First Guideline for Technologies
                            IST IP NOBEL "Next generation
                            Optical network for Broadband            ee787c6d-c81c-433e-a3ea-
                                 European Leadership"                       3ab4c127c833.doc




DWDM    Dense Wavelength Division Multiplexing
EAM     Electro-Absorption Modulator
EDFA    Erbium Doped Fibre Amplifier
EOP     Eye Opening
FEC     Forward Error-Correction
FFE     Feed-Forward Equaliser
FWM     Four Wave Mixing
GDR     Group-Delay Ripple
ISI     Inter-Symbol Interference
LH      Long Haul
LMN     Least Mean Square
MAN     Metropolitan Area Networks
MC      Monte Carlo
MLSE    Maximum Likelihood Sequence Estimation
MSE     Mean Square Error
MUX     Multiplexer
MZM     Mach-Zehnder Modulator
NRZ     Non-Return to Zero
OA      Optical Amplifier
OADM    Optical Add-Drop Multiplexer
OADN    Optical Add-Drop Node
OSNR    Optical Signal to Noise Ratio
PMD     Polarisation Mode Dispersion
PMDC    Polarisation Mode Dispersion Compensator
PRBS    Pseudo Random Bit Sequence
PSO     Particle Swarm Optimisation
ROADM   Reconfigurable Optical Add-Drop Multiplexer
Rx      Receiver
RZ      Return to Zero
SBS     Stimulated Brillouin Scattering
SOA     Semiconductor Optical Amplifier
SPM     Self-Phase Modulation
SRS     Stimulated Raman Scattering
SSMF    Standard Single-Mode Fibre




                                    Page 7 of 144
                                                                 First Guideline for Technologies
                                 IST IP NOBEL "Next generation
                                 Optical network for Broadband             ee787c6d-c81c-433e-a3ea-
                                      European Leadership"                        3ab4c127c833.doc




TDC          Tuneable Dispersion Compensation
Tx           Transmitter
UHL          Ultra Long Haul
VE           Viterbi Equaliser
WAN          Wide Area Network
XPM          Cross-Phase Modulation




    1.3.3    Definitions




    1.4     Document History

Version                    Date                      Authors                 Comment
1                          01.09.2004                Anders Djupsjöbacka     Initial document




    2       Document overview
The document is structured as follows.




                                         Page 8 of 144
                                                                  First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                           3ab4c127c833.doc




 3        Socio-economic and technical environment

 3.1      Technological progress
        Broadband access
        Broadband for all
        New services to generate new revenues



 3.2      Customer demands and market trends
        Internet anywhere, anytime
        Mobility
        Fast access
        Services on demand
        Convergence of networks
        Reduced OPEX
        Reliable and robust, dynamic, high capacity networks to deliver services



 3.3      Considered model networks
Models of a dynamic networks with definitions and specification taken also from other WPs.



 3.4      Technologies to enable future dynamic optical networks
Brief discussion of how technologies, components and subsystems can support optical
networks. Which novel technologies account for the market trends identified in 2.2 and how
do they do this.




                                        Page 9 of 144
                                                               First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband            ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                       3ab4c127c833.doc




 4        Current component developments
Detailed discussion of trends in the component development. Where are the foci?
Goals of recent component developments for WDM systems can be categorized in
     1. Increase of system performance
     2. New functions
     3. Lower costs
     4. Reduction of size / higher integration



 4.1      Technologies to increase system performance
Increase of data rates
Alternative modulation formats
Optical Compensators
Electrical or optical equalizers
Amplifiers




 4.2      Technologies to enable new functionalities
Reconfigurable OADMs, OXCs
Burst technologies
Optical 2R/3R
Lambda conversion



 4.3      Technologies to lower CAPEX or OPEX
Technologies which reduce OAM e.g.
        Remotely reconfigurable devices (tunable or reconfigurable OADMs, OXCs)
        Tunable transmitters and receivers
        Lambda converters
        Regenerators
OPM and failure root cause analysis




                                       Page 10 of 144
                                                               First Guideline for Technologies
                            IST IP NOBEL "Next generation
                            Optical network for Broadband               ee787c6d-c81c-433e-a3ea-
                                 European Leadership"                          3ab4c127c833.doc




Technologies to increase system performance have to reduce at least costs per bit




 4.4    Approaches to deliver products of higher integration
SMA approach
SFPs


 4.5    Technologies and components considered within
        deliverable D19

 List of components which are investigated here + motivation why focusing on those




                                     Page 11 of 144
                                               First Guideline for Technologies
               IST IP NOBEL "Next generation
               Optical network for Broadband            ee787c6d-c81c-433e-a3ea-
                    European Leadership"                       3ab4c127c833.doc




5   Transmitter/receiver and alternative
    modulation formats (DQPSK etc.)




                       Page 12 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                              3ab4c127c833.doc




 6       Tunable Laser
Tunable lasers are a basic building block for next generation optical network. An
introduction on tunable laser technology is provided in Nobel deliverable D3, Section 3.2.1.


Tunable Lasers Requirements
Transport network characteristics in terms of distance (access, metro, long haul), channel
spacing, (100, 50, and 25 GHz), and data rate (2.5, 10, and 40 Gbps) determine the
required optical performances. In addition to the transport applications, optical add/drop
multiplexing and switching applications to enable dynamic rerouting drive a different set of
performance requirements that are largely dependent on output power, tuning range and
tuning time.
Considering modulation, other parameter can be of interest. For directly modulated lasers,
the modulation bit-rate: the current modulation in the laser gain section involves a
frequency chirp or line broadening, so transmission is limited at 2.5 Gb/s signals over
metro-application distances. Another aspect that should be considered is the possibility of
the technological platform to integrate modulators, so modulator extinction ratio and chirp
are main parameters.
The following table lists the main characteristics that will be considered.




                                       Page 13 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




   Parameter         Symbol     Unit                         Description
Tuning                 BT        nm      Bandwidth of operation over which the laser can
Bandwidth                                be set over the ITU-T channel grid.
Channel Spacing               GHz      Supported channels grid
Tuning Time            TT        ms      Time needed to switch from one wavelength to
                                         another
Maximum Output         Po        mW      The fiber-coupled optical output power of the laser
Power
Optical power          Pv        dB      Fluctuation, over channels and operating
variation                                temperature, of the optical output power
Wavelength            Wa        GHz      It is the allowable offset from the ITU grid after the
Stability                                tuning. Depending on the technology, the target
                                         stability can be achieved with an embedded
                                         wavelength-locker
Maximum                Fw       GHz      For Single-Longitudinal Mode (SLM) sources, the
Spectral Width                           spectral width is defined as the full width of the
                                         largest spectral peak, measured 20 dB down from
                                         the maximum amplitude of the peak.
Side Mode            SMSR        dB      The Side Mode Suppression Ratio (SMSR) is
Suppression                              defined as the ratio of the largest peak of the total
Ratio                                    source spectrum to the second largest peak
Relative Intensity    RIN       dB/Hz    A measurement of the ratio between the noise
Noise                                    level at a particular modulation frequency to the
                                         average power of the signal
Polarization           Pe        dB      The ratio of two polarization states (TM mode and
Extinction Ratio                         TE mode) at output.
Optical Isolation      Oi        dB      Ratio of the back reflected optical output power
Maximum                Pe        W       Maximum electrical power consumption
Dissipated
Power


Another aspect to be considered is the non-traffic affecting “dark-tuning”, meaning that
during the tuning process the laser shall avoid to emit on wavelengths different from the
selected one. In case the technology inherently prevents this behaviour, a shutter must be
added.
In the assessment of existing or proposed technology, we will consider the following targets
(many of them being provisional values, to be modified according to the results of specific
Nobel activities in other WPs).




                                        Page 14 of 144
                                                                 First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                           3ab4c127c833.doc




   Parameter         Symbol     Unit                  Provisional Requirements
Tuning                 BT        nm      Full C or L-band
Bandwidth
Channel Spacing               GHz      50 GHz
Tuning Time            TT        ms      10 milliseconds
                                         (50 ns?)
Maximum Output         Po        mW      2-20 mW
Power
Optical power          Pv        dB      +/- 0.5 dB
variation
Wavelength             Wa       GHz      +/- 2.5 GHz
Stability
Maximum                Fw       GHz      5 MHz
Spectral Width
Side Mode            SMSR        dB      45
Suppression
Ratio
Relative Intensity    RIN       dB/Hz    -135
Noise
Polarization           Pe        dB      20
Extinction Ratio
Optical Isolation      Oi        dB      20
Maximum                Pe        W       5
Dissipated
Power


Among the requirements elaborated within Nobel activities, the characteristics required to
operate in an Optical Burst Switching scenario (analyzed by WP3) are considered. First
discussion of tunable laser requirements and experimental demonstrations to date can be
found in Deliverable D4, chapter 5.3.5 (“Fast tunable lasers: an important technology for
more flexibility”).


Tunable lasers could also reduce the need for large inventories of WDM line-
cards/modules for service providers and manufacturer.
An analysis of cost level with respect to tunability range has been proposed for savings
evaluation arising from tunability from a manufacturer point of view, basing on a generic
operational cost model1.
As represented in Figure 7-10 (representing the relative transmitter price, with respect to
single wavelength case, to achieve cost equivalence), the analysis outcomes showed that
tunability inventory savings would generally be most important for build-to-inventory (line




                                        Page 15 of 144
                                                                         First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                      ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                 3ab4c127c833.doc




with white triangles in Figure 7-10) of line cards (high overall unit price). Conversely,
tunability can enable little operational savings when inventory is small, like in build-to-order
or only few favored channels stock (line with black squares in Figure 7-10). This is an
example focused on metro applications and do not consider possible specific ‘channel
population strategies’ in the DWDM system, but gives an information about cost levels to
be achieved with respect to single wavelength lasers.




Figure 6-1 – Cost equivalence vs. tunability (White triangle: build-to-inventory. Black squares: build-
                                                        1
                                              to-order)


Tunable Lasers overview
Tunable laser technologies fall into the following main categories:
            o   External cavity lasers (ECL)
            o   Vertical-Cavity Surface-Emitting Lasers (VCSEL)
            o   Distributed Feedback Lasers (DFB)
            o   Distributed Bragg Reflectors (DBR) and multi-section DBRs
            o   Narrowly tunable lasers combination (Laser Array)
All but the VCSEL are based on edge-emitting devices, which emit light at the substrate
edges rather than at the surface of the laser diode chip. Vertical-cavity structures do the
opposite.
A laser’s wavelength is determined by its optical cavity. Besides the characteristics of each
technology, tunable lasers can emit on different wavelengths thanks to the ability to modify
the parameters of the resonant cavity. The resonant wavelength is actually determined by
the cavity length (mirrors distance) and by the speed of light within the medium that fills the
cavity (determined by the effective refractive index).
A general structure for a fixed wavelength laser can be represented in Figure 6-2, where
the resonant cavity between a couple of mirrors (with mirror-2 partially reflective) includes a




                                          Page 16 of 144
                                                                         First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                  3ab4c127c833.doc




gain media and a mode selection filter (enabling resonation of just one among the possible
cavity modes).
                                                            Mode
                                Gain                      selection
                                                            filter
                                                                            Output




                Mirror-1                                              Mirror-2


                   Figure 6-2 – General structure of a fixed wavelength laser

The operating wavelength of a semiconductor laser can be modified varying the cavity
length or changing the refraction index of the propagating media. There are several
methods to modify these parameters (mechanically, with MEMS, or via thermal effects or
via current injection, as can be seen in the following paragraphs).
For semiconductor lasers, there are three general wavelength tuning method:
           o   Carrier injection (or free carrier plasma effect)
           o   Quantum confined Stark effect (QCSE)
           o   Temperature tuning.
Tuning performance comparison of these mechanisms is shown in the following table2,
reporting for each tuning method some significant parameters: refractive index change n,
confinement factor , wavelength change l, phase-amplitude coupling factor H, level of
heat generation, and implementation difficulties.




Carrier injection is most widely used for tunable semiconductor lasers, due to broadest
tunability. Temperature tuning has to be anyway considered since current injection
determines temperature variations that affect the tuning (towards a reduction in tuning
efficiency, due to different sign of wavelength change).
Figure 6-3 represents the generic structure of a tunable laser, and the possible methods to
achieve wavelength tuning.




                                       Page 17 of 144
                                                                                  First Guideline for Technologies
                                  IST IP NOBEL "Next generation
                                  Optical network for Broadband                             ee787c6d-c81c-433e-a3ea-
                                       European Leadership"                                        3ab4c127c833.doc




                                                          Cavity     Mode
                                    Gain                  phase    selection
                                                                     filter
                                                                                       Output




                 Mirror-1                                                      Mirror-2


                            Figure 6-3 – General structure of a tunable laser

In the technological overview are included tunable laser components (typically in butterfly
packages) and modules (including electronic control functions).



DFB (Distributed Feedback) Tunable Lasers
DFB lasers incorporate a grating directly into the laser chip itself (Figure 6-4), usually along
the length of the active layer: the grating (acting as a mode selection filter) reflects a single
wavelength back into the cavity, forcing a single resonant mode within the laser, and
producing a stable, very narrow bandwidth output.
DFB emitted optical power can be 20 mW and the lasing action in the DFB also stabilizes
the carrier density in the device, leading to little wavelength drift over time (typically a 0.1
nm shift over 25 years) and enabling operation at 25 GHz channel spacing. Other DFB
typical characteristics are narrow linewidth and high optical purity (high side-mode
suppression ratio).
DFB is tunable in terms of current and temperature and tuning rate of the order of 0.1Å /
mA and 1Å/ °C respectively. DFB lasers are tuned by controlling the temperature of the
laser diode cavity. Because a large temperature difference is required to tune across only a
few nanometers, the tuning range of a single DFB laser cavity is limited to a small range of
wavelengths, typically under 5 nm. The typical tuning speed of a DFB laser is of several
seconds and the typical output power can reach 20 mW.
This laser is well suited for production in large volumes (the manufacturing process is
established) but the tuning range is narrow and to maintain optical performance over a
wide temperature range can be challenging.
                    Bragg reflector integrated with
                            active section


                         Active section
                                             HR coating                   AR coating




       Figure 6-4 – In a DFB laser the diffraction grating is integrated into the active section

DFB lasers are not widely tunable and they will not be further investigated in this
document, but nevertheless represent a common benchmark in terms of performances,
reliability and cost. Thermally tunable (typically over 8 channels at 50 GHz) DFB lasers with
the possibility to be directly modulated at 2.5 Gb/s are currently proposed on the market.




                                             Page 18 of 144
                                                                                   First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                               ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                          3ab4c127c833.doc




DBR Tunable Lasers
A variation of the DFB laser is the distributed Bragg reflector (DBR) laser. It operates in a
similar manner except that the grating, instead of being etched into the gain medium, is
positioned outside the active region of the cavity (Figure 6-5), also simplifying the epitaxial
process. Lasing occurs between two grating mirrors or between a grating mirror and a
cleaved facet of the semiconductor.
                                         Bragg reflector




                            HR coating                           AR coating
                                               Active section


             Figure 6-5 – In a DBR laser the grating is contained in a separate section

Tunable DBR lasers (Figure 6-6) are made up of a gain section, a mirror (grating) section
(for coarse tuning), and a phase section, the last of which creates an adjustable phase shift
between the gain material and the reflector (to align cavity mode with the reflection peak,
for fine tuning). Tuning is accomplished by injecting current into the phase and mirror
sections, which changes the carrier density in those sections, thereby changing their
refractive index (temperature can also be used to control refractive index changes, with
lower tuning speed). Thus, at least three control parameters have to be managed,
increasing the complexity of the system; moreover, the refractive index to current relation
changes with time (due to p–n junctions degradation, a certain current correspond to a
smaller carrier density).
                                 Bragg reflector           Phase section




                           HR coating                                 AR coating
                                                    Active section


                        Figure 6-6 – Tunable DBR laser general structure

The tuning range in a standard DBR laser seldom exceeds about 10 nm and it will not be
further covered in this document. Wider tuning can be achieved adding other sections
besides gain and phase sections and the various possible solutions are described in
related paragraphs below.
Being based on electrical effects, tuning speed is much faster than DFB, while optical
output power of DBR is generally lower than for DFB lasers.


Laser Array Tunable Lasers
The DFB thermal tuning range can be expanded by having an array of lasers of different
wavelengths integrated on the same chip.
DFB selectable arrays operate selecting the DFB array element for coarse tuning and then
exploiting temperature tuning for fine cavity mode tuning.




                                             Page 19 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                3ab4c127c833.doc




Common approaches to implement the coarse selection in DFB array are based on
integrated on-chip combiners or on off-chip MEMS deflectors able to route the proper beam
on the laser output.
The advantages of the on-chip combiner approach are mainly the reliability and spectral
characteristics that are basically the same as fixed wavelength sources. Disadvantages are
relevant to the trade-off between power and tuning range (sometimes a SOA is added to
counterbalance the combiner losses, that increment with the number of lasers), reduced
yield and large real estate requirements.
MEMS-based devices can improve optical output power and decrease chip size, but
introduce an element that can significantly affect reliability.

SANTUR (DFB array)
Santur approach is based on a DFB array and an external MEMS tilt mirror is used to
select the appropriate DFB 3.
Figure 6-7 depicts the structure of a device able to tune over 33 nm in C-band. The DFB
array chip contains 12 lasers with a wavelength spacing of 2.8nm. The mirror (only a small
deflection is needed since the array is closely spaced) is placed at the focal plane of the
collimating lens, and thus corrects for the spatial variation of the generated beam and
delivers to the fiber nearly the full output power of the laser.
The package is simplified (low cost passive alignment) since fine alignment is done
electronically with the tilt mirror.




  Figure 6-7 – Schematic of tunable laser package. The MEMS tilt mirror in the focal plane of the
    collimating lens selects one laser from the DFB array and allows for electronic fine tuning of
                                                        3
                                             alignment.

The module can provide 20mW power over the tuning range (Figure 6-8 shows
superimposed spectra at 50GHz ITU channels). Tuning time is typically a second between
channels, with the value limited by the TEC cooling capacity. The wavelength is locked on
to the ITU grid with the accuracy of the external 50GHz spaced wavelength locker.




                                         Page 20 of 144
                                                                  First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                            3ab4c127c833.doc




                                                                                            3
 Figure 6-8 – Superimposed spectra of module of 84 x 50GHz ITU channels at 13dBm or 20mW.

Control electronic manages three control loops. MEMs voltages are set to their calibrated
values, but are continually optimized to maximize the fiber-coupled power (as measured by
the wavelength locker, that also controls the temperature to stay on the grid), in case
adjusting laser current to equilibrate the optical power. The feedback loop maintains the
optimum alignment and compensates for possible mechanical drift or creep.
During a wavelength-switching event, the MEMs mirror moves to an extreme position to
blank the output by about 50dB while the new temperature and current values stabilize
(thus acting as a built-in shutter/VOA). The MEMs is then unblanked and the locker can
provide fine wavelength control. The MEMs mirror can also be detuned during normal
operation to provide an additional VOA functionality.
The MEMs mirror is fabricated using bulk silicon micromachining and measures about 1.5
mm on a side and is coated with gold for high reflectivity at 1.55mm. The electrostatic
mirror deflection is obtained by applying a voltage between the mirror surface and one of
the pads beneath the mirror.
Shock and vibration have no effect on the wavelength stability of the laser, but only cause
minor amplitude modulation of fiber coupled power.
The chip size is similar to that of fixed wavelength DFBs, and it contains no additional
processing steps (just different masks that result in 12 lasers rather than one). The lasers
share the same gain medium, and only vary in grating pitch, which is defined through direct
write e-beam lithography.
The total chip size is 1 mm long, while the laser ridges are only 500 microns4.




                                      Page 21 of 144
                                                                           First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                         ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                    3ab4c127c833.doc




                                                                       4
                            Figure 6-9 – Fully processed laser chip.

Further improvements have been studied in order to have a less expensive and more
compact solution, substituting the external wavelength locker with an integrated
wavelength locker and a quad detector for C-band operation at a channel spacing of
25GHz5.
As represented in Figure 6-10 beamsplitters are used to pick off part of the collimated
beam and to direct a portion of it through a solid etalon to a standard photodetector, while
the remainder is transmitted to a quadrant-type photodetector (used as a measure of the
wavelength error, adjusted changing the DFB array temperature).
The quadrant detector is used to measure the position of the beam reflected from the
mirror by determining the relative power incident on each of the four quadrants. The
position signal is then used by the control electronics to fine-tune the mirror position.
The etalon is held at a constant temperature during operation using a separate TEC with
respect to the DFB array; the possibility to adjust the etalon temperature also allows the
etalon to be passively placed during assembly.




                                                                              5
                       Figure 6-10 – Schematic of tunable laser package.

Figure 6-11 shows the frequency error and the deviation of the output power from the
power set point that results when the module was switched randomly among 16 channels
1000 times. Frequency accuracy of +/ - 1.0GHz and power stability of +/-0.5dBm are
routinely achieved.



                                       Page 22 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                  3ab4c127c833.doc




 Figure 6-11 – Measured frequency error (left) and output power error (right) when the module was
                                                                                     5
              randomly switched 1000 times among different channels in the C-band.

An integrated solution including a 10 Gb/s LiNbO3 modulator and a liquid-crystal-
based VOA (Figure 6-12) has been presented6.
VOA enables the output powers of the individual transmitters to be varied in order to
change the spectral power density for gain tilt control of the EDFA. The dynamic range of
modulated output power is typically 15 dB.
This VOA, in conjunction with the modulator, can be used for output shuttering while the
laser is tuned. It can also be used to slowly bring up or down the transmitter so the network
does not experience sudden changes in the spectral power density.




                                                                               6
                       Figure 6-12 – Structure of an integrated transmitter.




                                         Page 23 of 144
                                                                     First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                              3ab4c127c833.doc




Figure 6-13 shows that the variation in BER over the C-band is very small and limited by
test capability.




                                                                                                  6
  Figure 6-13 – Bit-error-rate (BER) in back-to-back over several channels spanning the C-band.

Santur proposes TL-2000 series, with models able to tune over C or L bands and an output
power of 10 or 20 mW, with integrated wavelength locker for 25 GHz operation. The
product is a module OIF MSA compliant.
In the following table, specifications of TL-2020-C (20 mw, C-band) are reported7.

   Parameter         Symbol       Unit                          TL-2020-C
Tuning                  BT         nm     36
Bandwidth
Channel Spacing                 GHz     25 (50 typ.)
Tuning Time             TT         ms     2000
Maximum Output          Po        mW      20
Power
Optical power           Pv         dB     +/- 0.3 over temperature
variation
                                          +/- 0.3 over wavelength
Wavelength              Wa        GHz     +/- 1.5
Stability
Maximum                 Fw        MHz     10 (FWHM)
Spectral Width
Side Mode             SMSR         dB     > 40
Suppression
Ratio
Relative Intensity     RIN       dB/Hz    -135 (20 MHz to 10 GHz)




                                         Page 24 of 144
                                                                        First Guideline for Technologies
                                 IST IP NOBEL "Next generation
                                 Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                      European Leadership"                              3ab4c127c833.doc




Noise
Polarization              Pe         dB      20
Extinction Ratio
Optical Isolation          Oi        dB      30
Maximum                   Pe          W      7.5
Dissipated
Power



AGILENT (DBR array)
A monolithic tunable laser source based on a selectable DBR array has been
demonstrated8. Compared to DFB selectable array, this tunable source is tuned by injected
currents and it does not imply any thermal tuning (thus achieving reduced switching time).
The BDR array is monolithically integrated (Figure 6-14) with a Planar Integrated Circuit,
capable of routing the power to a MMI (Multi-Mode Interference) coupler and a SOA
(Semiconductor Optical Amplifier).




                                                                                     8
                    Figure 6-14 – Schematic drawing of the wide tunable laser chip

The device can be tuned over 40 nm with 20 mW of output power exfacet and has a power
consumption of 250 mW.
The device consists of a 4 DBR array: the 2-sections DBRs have been designed to get
high power, high extinction ratio and wide tuning range (>10nm). The grating section has
been designed to get maximum power from its facet, high SMSR (>30dB), acceptably low
propagation losses, wide tuning (>10nm).
The SOA (recovering the internal losses and boosting the optical power up to 20 mW) has
been designed to achieve a high saturation power over the C-band and to recover all the
optical losses, together with a bent output waveguide, essential to reduce the AR
requirements.
Figure 6-15 shows the spectral behaviour, with more than 40 dB SMSR.




                                           Page 25 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                3ab4c127c833.doc




Figure 6-16 shows the Relative Intensity Noise (RIN peak in 15 GHz bandwidth)
characteristics for all the channels, which always results better than -135 dB/Hz.
The device optical power (measured with respect to the various parameters such as DBR
currents and SOA injections) is up to +13 dBm ex facet, leading to + 8 dBm in fiber (a
coupled power close to 10 dBm is achievable by carefully optimising the fiber output
angle).




Figure 6-15 –Superimposition of the 41 nm total tuning range achieved by the DBR selectable array.
                                                                                      8
            The colours underline the tuning achieved by one of the DBR array (10 nm)




                                                                                                     8
 Figure 6-16 – Relative Intensity Noise RIN for all the channels, at various SOA injected currents

The DBR array (only 1670 m long and 400 m wide) is fabricated on InP substrate. DBR
active section and SOA are based on a Multi Quantum Well stack. The passive quaternary
waveguides, is a InGaAsP bulk waveguide, the grating is written by Electron Beam
Lithography and defined by RIE in another InGaAsP bulk layer. InP:Fe blocking layer has
been re-grown to define the Semi-Insulating Buried Heterostructure (SIBH) surrounding the
mesas.




                                         Page 26 of 144
                                                                    First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                      ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                                 3ab4c127c833.doc




                                                                            8
                    Figure 6-17 – SEM picture of the tunable laser source


NEC (DFB array)
As an example of typical architecture for DFB laser array with integrated combiner the one
proposed by NEC can be discussed9. It is based on a DFB microarray, a multi-mode
interference (MMI) optical coupler, and a semiconductor optical amplifier (SOA) and the
resulting device is able to cover the entire S-, C- and L-bands on a 100-GHz ITU-T grid by
six eight-DFB-array (Figure 6-18). The tuning ranges (∆λ) of each device are distributed
from 15 nm to 20 nm.




                                                                                  9
               Figure 6-18 –Schematic structure of fabricated eight-array-WSLs.

Output power of more than 9 mW was obtained for all the six devices for all temperatures
in the range. The SOA gain (about 10 to 12 dB) is enough to compensate for the 1 × 8 MMI
splitting loss (9 dB) and waveguide propagation loss (1.2 dB), while the estimated fiber
coupling loss is about 3.4 dB.




                                       Page 27 of 144
                                                                      First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                                3ab4c127c833.doc




Figure 6-19 shows the superimposed lasing spectra of the six arrays (a total of 135
channels over the entire S-, C-, and L-bands). Each wavelength can be tuned to 100-GHz
ITU-T grids by selecting a DFB-LD and varying the temperature by 25 K.
DFB array offer stable and reliable wavelength properties. All devices show stable single-
mode oscillations (having an SMSR of more than 42 dB under the entire tuning wavelength
range).




                                                                      9
                          Figure 6-19 – Lasing spectra of six WSLs.

The integration structure uses a selectively-grown microarray active waveguides directly
connected to dry-etched passive waveguides.
The device fabrication process includes four growth steps and two dry etching steps. Since
both the active and the passive-waveguides have a common MQW waveguide, low-loss
and low-reflective couplings are implemented at the interface (Figure 6-20).




                                                                                   9
             Figure 6-20 –SEM photograph of integrated active-passive waveguide.




                                      Page 28 of 144
                                                                     First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                               3ab4c127c833.doc




NTT (DBR array)
The DBR laser is a suitable candidate for applications demanding fast switching. Inherently
mode-hop-free operation is also a requirement (normally this also enable stable
characteristics when direct modulation is adopted). These characteristics have been
shown10 together with a wavelength selectable 4-ch DBR laser array with a multi-mode-
interferometer (MMI) coupler and a booster semiconductor optical amplifier (SOA),
achieving a modehope-free wavelength tuning over 12.45 nm.
The device (Figure 6-21) consists of a short active region and DBR regions on both sides
of the active region. Both DBR regions are electrically connected for wavelength tuning.




                                                                                                  10
 Figure 6-21 – Schematic structure of the mode-hop-free DBR laser with the short active region.

Since the active region length is reduced to less than 50 µm (with respect to conventional
tunable DBR laser with several-hundred-micrometer-long active region), the continuous
tuning range without mode-hop is expanded (Figure 6-22).




                                        Page 29 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




 Figure 6-22 – Active region length dependence of mode-hop-free tuning range. Experimental data
                                                                               10
                   with active region lengths of 50 and 100 µm are also shown.

The 2.5 Gbit/s direct modulation was also demonstrate, with error-free operation under
tuning (High speed switching time of few nano-seconds) for back-to-back and 20-km
transmission over standard single mode fiber.
Using these mode-hop-free DBR lasers, a monolithically integrated wavelength-selectable
4-ch DBR laser array was also fabricated (Figure 6-23). The lasing wavelength is tuned by
selecting one DBR laser and then by current tuning to the DBR regions.
The averaged value of the wavelength difference between each laser is 2.5 nm. The mode-
hop-free tuning in each DBR laser was measured to be more than 4 nm at the tuning
current injection of up to 100 mA.




Figure 6-23 –Schematic structure of the wavelength-selectable 4-ch DBR laser array with MMI and
                                                  10
                                            SOA.




                                       Page 30 of 144
                                                                First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband              ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                         3ab4c127c833.doc




This device was fabricated using five MOVPE growth steps. A bulk layer was used for large
optical confinement in the active region, and large coupling coefficient of corrugation
grating κ was used for high reflection in the DBR regions (formed by wet-etching). The BH
mesa (buried by p- and n- type current blocking layers) was formed by CH4 +H2 dry
etching.
Currently there is not evidence of publicly proposed products based on this technology.

QDI DFB array
QDI proposes “WSL® Tunable DFB Laser Array Module With Wavelength Locker”, a
product based on temperature-tuned wavelength selectable DFB laser arrays. The 16-
element DFB laser array chip is monolithically integrated with a combiner and a
semiconductor optical amplifier (SOA). The module
WSL® Tunable DFB Laser Array Module With Wavelength Locker
In the following table, specifications of WSL® Tunable DFB Laser Array Module With
Wavelength Locker (10 mw, C-band) are reported11.

   Parameter         Symbol     Unit                           WSL
Tuning                 BT        nm      32
Bandwidth
Channel Spacing               GHz      50
Tuning Time            TT        ms      -
Maximum Output         Po        mW      10
Power
Optical power          Pv        dB      -
variation
Wavelength            Wa        GHz      -
Stability
Maximum                Fw       MHz      15
Spectral Width
Side Mode            SMSR        dB      -
Suppression
Ratio
Relative Intensity    RIN       dB/Hz    135
Noise
Polarization           Pe        dB      -
Extinction Ratio
Optical Isolation      Oi        dB      -
Maximum                Pe        W       -
Dissipated
Power




                                        Page 31 of 144
                                                                           First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                  3ab4c127c833.doc




Multi-Section DBR Tunable Lasers
In order to improve tunable DBR performances different solutions, based on the concept of
incorporating additional elements (control/gain sections) to the basic tunable DBR, have
been developed.
One among them is the Sampled Grating DBR (SG-DBR).
SG-DBR (Figure 6-24) uses two gratings (placed at the opposite ends of the gain section)
with a slightly different step, thus obtaining two wavelength combs, with a slight offset.
During tuning (obtained varying the current flowing into the front and rear gratings and
phase section), the gratings are adjusted so that the resonant wavelengths of each grating
are matched. The difference in blank spacings of each grating means that only a single
wavelength can be tuned at any one time. Due to this arrangement (exploiting Vernier-like
effect of reflection peaks of the two grating sections) a small tuning of the combs results in
a significant change in the resonant lambda and thus in a wider tuning range.
SG-DBR lasers are a special case of a more general structure, (super structure grating
SSG-DBR), where the front and rear gratings can be sampled with a modulation function
(such as a linearly chirped grating), obtaining different shapes of envelope of the reflectivity
peaks (the reflection envelope shape depends on Fourier components of the modulating
function).
                                          Phase section   Front Bragg reflector




                   HR coating                                        AR coating
                                       Active section


                       Figure 6-24 – Tunable SG-DBR laser general structure

Tuning is not continuous (Vernier-like effect means that wavelength will jump in step, so
quasi-continuous wavelength tuning can be achieved using the gain and mirrors sections,
while phase section provides fine tuning) and multiple sections are involved, so requiring a
more complex control than a standard DBR or DFB laser.
The output power is typically less than a standard DBR and is about 2 mW, due to more
passive sections in the cavity. Moreover current-injection-based refractive index tuning
produces an increase in absorption that results in a power variation of several dBs across
the tuning range (so large variations in gain current would be needed across channels to
maintain a constant channel-to-channel power). For these reasons SOA integration
(enabling higher output power, constant power level for all channels and blank output
power during the tuning process) is a possible choice, even at the cost of an increased
complexity (Figure 6-25).
Further integration with EA modulator is in general possible, taking into account increased
complexity.




                                         Page 32 of 144
                                                                                   First Guideline for Technologies
                                    IST IP NOBEL "Next generation
                                    Optical network for Broadband                                ee787c6d-c81c-433e-a3ea-
                                         European Leadership"                                           3ab4c127c833.doc




                                           Phase section Front Bragg reflector




             HR coating                                                             AR coating
                                               Active section            SOA


                             Figure 6-25 – Tunable SG-DBR laser with SOA

Switching speeds is inherently fast (tens of nanoseconds), but due to involvement of
control algorithms in the tuning process, the achievable switching time is of some
milliseconds. Considering fast switching requirements for OBS applications, limits in
switching time due to thermal drift should also be considered12.
Manufacturing process of SG-DBR lasers is similar to that of DBR lasers. This technology
is also suitable for EA modulator and SOA integration, with manufacturing process to be
yet improved, in terms of resulting yield.
Another version of multisection DBR is the Grating-assisted Coupler with Sampled
Reflector (GCSR) laser, which contains four sections (gain, Bragg reflector, coupling and
phase-correction) and is tuned using three currents. The current-controlled waveguide
coupler acts as a coarse tuner to deliver a narrow range of wavelengths from the
modulated Bragg reflector (providing a comb of peaks and which is itself current-controlled
to provide a level of selection), to the phase correction section (also current-controlled),
which acts as a fine tuning section (Figure 6-26). The concept is to match the reflection
peak spacing of the sampled grating to the filter width of grating assisted coupler. GCSRs
operate over a wide tuning range, on the order of 40 nm13, but their complex electronic
controls can limit switching speed (that is inherently around 20 ns). As for other
multisection lasers, power output is low (around 2 mW) and can be increased, at the
expense of tuning range, by eliminating the coarse tuning section.
                          Rear section Phase section               Active section




                     HR coating                                                AR coating
                                  Bragg reflector   Wavelength coupler


                                     Figure 6-26 – Tunable GCSR laser

In the following, multi-sections DBR products are investigated.



AGILITY (SG-DBR)
Agility proposes a family of widely tunable sources based on SG-DBR technology, with
integrated SOA: C and L-band tunable products supporting 50 and 25 GHz grid, with or
without integrated modulator for 2.5 Gb/s operation and integrated transponder modules.
The basic features of the full-band tunable SG-DBR with SOA are presented14.
The SG-DBR laser (Figure 6-27) gain section contains an active-layer embedded within a
InP-based semiconductor waveguide structure (current injection controls output power).



                                               Page 33 of 144
                                                                  First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                               3ab4c127c833.doc




Front and back DBR mirror sections are formed by sampled gratings, which have been
etched into passive waveguides (injected carriers induce a change in refractive index, thus
wavelength tuning). The phase section does not contain gratings or active material and
injected carriers induce change in refractive index used for fine tuning. The amplifier
section is located outside the front mirror: current injection controls output power.




                                                                          14
                    Figure 6-27 – Agility Tunable SG-DBR laser with SOA

Sampled gratings are obtained removing (blanking) the grating in a periodic fashion along
the length of a mirror section. This results in reflectance spectrum with a periodic comb of
high reflectance peaks rather than the single peak of a non-sampled DBR. The lasing
wavelength will be that of the cavity mode coinciding with the front and back mirror
reflectivity peaks with highest spectral overlap (Figure 6-28).
Coarse tuning is obtained with a differential tuning between front and back mirror sections,
allowing a new set of front and back mirror peaks to come into alignment (Vernier-like
effect). Current injection in the phase section tunes all cavity modes simultaneously, thus
equal tuning of both mirrors along with cavity modes results in a shift in lasing wavelength
proportional to index shift (fine tuning).
Coordinated current injection into all three sections results in quasi-continuous wavelength
coverage. Lasing on an absolute discrete frequency grid is achieved through calibration at
time of assembly.




                                       Page 34 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




Figure 6-28 – Calculated sampled-grating mirror reflectance spectra; in this example, lasing should
                                                                                14
                occur on the front and back mirror peaks aligned near 1555 nm

Power control with the SOA effectively decouples output power from other lasing
properties. Integrated SOA for power control allows the gain section current to be held
constant across channels (Figure 6-29), allowing the laser to be biased far enough above
threshold to ensure adequate spectral properties and also eliminating a source of parasitic
thermally-induced wavelength tuning.
SOA supplies a gain of only 3-8 dB at typical output powers and is operated in saturation,
so it has very little detrimental effect on noise or side-mode suppression. SOA also allows
for additional functionality, such as variable optical attenuation (VOA) over a wide (>15 dB)
output power range, as well as beam blanking during channel switching.




  Figure 6-29 – Fiber-coupled output power vs. wavelength for an SG-DBR-SOA laser at constant
                                                                      14
                            amplifier and gain current of 150mA each.




                                         Page 35 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                               3ab4c127c833.doc




The laser is calibrated on a 25 GHz grid from 1526.827 nm (196350 GHz) to 1568.773 nm
(191100 GHz) at a nominal power of +13.5 dBm. Figure 6-30 shows frequency deviation
and power variation vs. frequency for 211 channels. Frequency error is well within 1 GHz.
Channel-to-channel power variation of ~+0.1 dB is also achieved through power control on
the SOA current.




 Figure 6-30 – Frequency deviation from ITU grid (left) and power variation across channels (right),
                                                                                    14
              for a module calibrated on a 25 GHz frequency grid for 211 channels.

Figure 6-31 shows side mode suppression ratio (measured for a resolution bandwidth of
0.1 nm), that is better than 41.27 dB across all channels, with a peak of 51.5 dB. Typically,
the highest side mode occurs at an adjacent set of mirror peaks, approximately 5 nm
longer or shorter than the lasing wavelength.




                                                                                                   14
  Figure 6-31 – Side mode suppression ratio vs. laser optical frequency for a calibrated module.

Figure 6-32 shows that Relative Iintensity Noise (RIN) better than -141.8 dB/Hz is achieved
in the tuning bandwidth. Noise is strongly correlated with mirror and phase section tuning-
induced absorption.




                                         Page 36 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




   Figure 6-32 – Relative intensity noise (RIN) vs. laser frequency, averaged over 0.3 to 10 GHz
                                                          14
                                           spectral range

Figure 6-33 shows the laser linewidth, varying between 800kHz and 5 MHz across the
tuning range (correlated with tuning current).




Figure 6-33 – Linewidth vs. laser optical frequency obtained from frequency noise spectra averaged
                                                                  14
                                  over 1 to 2 GHz spectral range.

Since the SG-DBR-SOA laser has ideally 3 terminal currents affecting wavelength (front
mirror, back mirror, phase) and two currents controlling power (gain and SOA), both
wavelength and power are in general over-determined. The set of operating currents to
achieve a particular wavelength is not unique, but instead spans one or more continuous
surfaces in current space. The calibration phase is carried out obtaining a mode map and
corresponding operating points measured at constant gain and SOA currents, and at a
constant phase current.




                                         Page 37 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




In order to guarantee the operation on ITU-T grid, characterization is performed under
automatic power control, so that phase current and SOA current are allowed to vary based
on feedback from an integrated wavelength locker.
Laser control, to ensure wavelength accuracy and reliable operation under environmental
or aging-induced shift in tuning performance over the lifetime of the device, must be very
accurate, especially for SG-DBR with SOA devices that have several control parameters.
Figure 6-34 shows a block diagram of the integrated module highlighting control elements
and parameters, all controlled by a digital signal processor, managing the device
according to a lookup table containing current settings for each section, along with other
channel-specific parameters, at each channel. Besides calibration, closed loops (for
example power measured from the wavelength locker reported back to amplifier section)
allow adjusting over device lifetime. Another control loop involving mode control (by
dithering a mirror current about its nominal set point while simultaneously sampling the
gain section voltage) ensures minimal wavelength deviation.




                                                                                            14
        Figure 6-34 – Schematic diagram of tunable laser module with control electronics.

It is important to determine how the increased complexity with respect to standard tunable
DBR affects reliability. DBR laser device life is limited by tuning section degradation15,
while for SGDBR lasers the tuning section lifetimes do not limit the overall device life16.
In DFB or standard DBR lasers a large source of the material defects that cause the device
to degrade in time is the grating structure where semiconductor materials have been
etched and regrown; moreover the current applied to the device is directly determining
degradation. SGDBR lasers have a low percentage of gratings due to their periodic
sampled structure and tuning current densities applied to the mirror and phase sections of
the SGDBR are much lower than DBR or DFB.
Results of accelerated life testing of SGDBR lasers show that the active and tuning failure
rates (no mode-hopping occurring over device lifetime) results in a low total wear out failure
rate that remains less than 150 FITs over the useful life of the device.
Agility proposes 3105/3106 and 3205/3206 (50 GHz spacing) products. Products with
integrated EAM (4235/5245, see below) and transponder (7100) are also proposed.
In the following table, specifications of 3205 are reported17.



                                        Page 38 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




   Parameter         Symbol     Unit                           3205
Tuning                 BT        nm      35 (from 1528.384 nm to 1563.863 nm C-band)
Bandwidth
                                         40 (from 1568.363 to 1608.329 nm L-band)
Channel Spacing               GHz      50
Tuning Time            TT        ms      10
Maximum Output         Po        mW      20 (typ.)
Power
                                         -30 dBm power-off tuning
Optical power          Pv        dB      +/- 0.3 over temperature
variation
                                         +/- 0.3 over wavelength
Wavelength            Wa        GHz      +/- 2.5
Stability
Maximum                Fw       MHz      10
Spectral Width
Side Mode            SMSR        dB      > 40
Suppression
Ratio
Relative Intensity    RIN       dB/Hz    < -135
Noise
Polarization           Pe        dB      -
Extinction Ratio
Optical Isolation      Oi        dB      -
Maximum                Pe        W       < 6.5
Dissipated
Power


Evaluations have been carried out on directly modulated SG-DBR to target low cost,
short reach applications18.
The device (Figure 6-35) is mounted on specially designed microwave carriers using
substrate able to minimize the effect of parasitic elements on the device response.
Tests of direct intensity modulation showed that the device can be direct modulated with
analog signals up to 6-GHz and can provide single-mode operation up to a 10-dB
extinction ratio. Tests of 2.5 Gb/s transmission over 75-km of an ordinary SMF were also
carried out.




                                        Page 39 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                              3ab4c127c833.doc




                                                                                               18
   Figure 6-35 – SG-DBR layout including interconnections for DC-bias and modulation signal.

Agility studied19 and now proposes SG-DBR with SOA and integrated modulator.
In principle, adding a modulator section to the SG-DBR with SOA (Figure 6-36) results
into an integrated solution (and the same chip can be operated as CW sources by keeping
the EAM in the on-state), but due to modulator bandwidth a widely tunable solution
requires to vary the modulator bias across the tuning range (main limitations of the EA-
modulator are inherent wavelength dependence of extinction ratio and chirp and trade-off
between these characteristics and insertion loss)20. Moreover this structure can be a viable
solution for 2.5 Gb/s on short/medium distances, but poses some limits for higher bit rate
and longer distances.




Figure 6-36 – Single-chip widely-tunable transmitter schematic showing a SG-DBR laser integrated
                                                           19
                                     with an SOA and EAM.

One possible solution to improve tunable laser performances in terms of chirp control is a
‘tandem EA’ modulator (Figure 6-37), employing a phase section (driven with an inverted




                                        Page 40 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                              3ab4c127c833.doc




data signal) in addition to the electro-absorption section to compensate the positive chirp
usually observed.




Figure 6-37 – SG-DBR integrated with a two-section tandem EAM including inverted data applied to
a phase modulator to compensate chirp. Amplitude modulator reverse biased for absorption; no bias
                                                          19
                                    applied to phase mod.

Another possible solution to control chirp involves replacing the EAM with a Mach-Zehnder
modulator (MZM), which seems a viable solution having solved integration difficulties due
to reflections. Monolithical integration (Figure 6-38) also results in small footprint and low
power dissipation.
MZM with dual drive allows a programmable chirp from +1 to –1 and error free
transmission over 80 km of standard fiber was demonstrated for all channels at 10Gb/s
using a negative chirp configuration.




                                                                                       19
        Figure 6-38 – SEM photo of SG-DBR integrated with a Mach-Zehnder modulator.

Basing on monolithically integrated SG-DBR-SOA-MZM transmitter structure a widely
tunable, 10 Gb/s transmitter, RF extinction ratio > 12 dB with less than 3 V modulation
voltage and negative chirp across a 40 nm tuning range (Figure 6-39) has been



                                        Page 41 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                              3ab4c127c833.doc




demonstrated21. Error-free transmission at 10 Gb/s has been demonstrated for 100 km of
standard single mode fiber.




   Figure 6-39 – RF ER and 3 dB alpha parameter in single-ended drive configuration across the
                                                      21
                                        tuning range.

Characterisation of phase noise has been carried out22. Laser phase noise is an important
parameter to be considered, also depending on the modulation format adopted. Anyway,
phase modulation causes additional intensity modulation (after transmission through fiber
due to fiber chromatic dispersion and nonlinearities). The efficiency of the phase
modulation-to-intensity modulation (PM-to-IM) conversion depends on frequency, and it is
quite important to identify how each frequency noise degrades the transmission
characteristics of lasers (measuring RIN after transmission through fiber and identifying
additional intensity noise caused by each frequency component of the phase noise).
Fixed-wavelength semiconductor lasers show white frequency noise, which results from
spontaneous emission and carrier-density fluctuations. Multi-section tunable laser diodes
exhibit (Figure 6-40), besides white frequency noise, which depends on gain current, at
frequencies between 500 MHz and 2 GHz, additional frequency noises, such as 1/f-like
frequency noise below 200 MHz, which does not depend on the gain current (1/f carrier
noise and injection-recombination shot noise, generated in the tuning sections, which are
biased below threshold).




                                        Page 42 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                               3ab4c127c833.doc




Figure 6-40 – Phase noise spectrum of a LDM in log scale from 1 MHz to 10 GHz for a gain current
                                                                                 22
             = 150 mA and an SOA current = 130mA. The wavelength is 1553.4 nm.

Figure 6-41 shows the RIN after 50 km fiber of the TLA (tunable laser assembly, consisting
of LDM plus current and temperature control circuits) for various fiber launch power. The
peaks and dips observed in the RIN spectrum result from the PM-to-IM conversion. The
tone observed at 11 GHz results from SBS, and its magnitude increases with fiber launch
power. Dithering reduces RIN after fiber at low frequency (below 200 MHz), caused by
SBS.




Figure 6-41 – RIN after 50km fiber of TLA for a fiber launch power of 7.4, 9.0, and 11.0 dBm. Data is
                                                                                        22
            averaged 32 times over frequency. The channel wavelength is 1551.32nm.




                                          Page 43 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                3ab4c127c833.doc




The peaks of RIN after fiber mainly depend on laser linewidth. RIN after fiber is determined
by the white frequency noise and is not affected by the non-white components. Therefore,
the transmission performance of the laser can be evaluated by the linewidth defined by the
white frequency noise (below 2 MHz for all the channels across the entire C-band as can
be seen from Figure 6-42).




Figure 6-42 – Linewidth of TLA defined by white noise level for all the channels across the entire C-
                                                              22
                                   band with 50GHz spacing.


BOOKHAM (DS-DBR)
The digital supermode DBR (DS-DBR) is fundamentally a sampled grating SG-DBR with a
different front-end reflector.
The front grating design is the key to the tuning mechanism. When activated electrically, its
chirped grating structure selects one of the supermode reflection peaks created by the rear
phase grating acting as a comb filter. The supermode that is selected depends on which
contact receives current, rather than on the magnitude of the current. This mechanism
results in excellent power uniformity, since there is very little current-induced absorption
within the device.
The DS-DBR eliminates the need to match the front and back gratings, so only two
selections (initial operating wavelength and any fine-tuning at the rear grating) are needed
instead of three.
A monolithic design integrating a digital supermode DBR (DS-DBR) laser with a
semiconductor optical amplifier (SOA) has been demonstrated23.
The surface-ridge InP laser is fabricated and packaged with conventional laser chip
processing (Figure 6-43). The front tuning section is a linearly chirped holographic Bragg
grating (improved modal discrimination over the bandwidth of 70nm) and the rear section




                                         Page 44 of 144
                                                                      First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                               3ab4c127c833.doc




has an e-beam written phase grating reflector that provides a sharp and flat comb
reflectance response.




                                                                 23
                              Figure 6-43 – DS-DBR laser chip.

In free running conditions (no tuning) the basic DS-DBR structure emits 40mW ex-facet
powers, so the integrated SOA run at high input power and low gain (~3-6dB), obtaining
powers in excess of 100mW. In order to prevent significant injection of an amplified return
wave into the DS-DBR, an angled output facet was used in conjunction with an anti-
reflective coating.
DS-DBR tuning is similar to that of an integrated set of 3-section DBR lasers. Narrowband
wavelength tuning is achieved by scanning the rear grating current. Continuous tuning can
then be arranged by using the phase section for longitudinal mode tracking.
Figure 6-44 shows the tuning characteristic, with five supermodes, while Figure 6-45 shows
an example of the 3-section like rear-phase tuning plane.
The thick sloping line in Figure 6-44 corresponds to the zero phase current axis of Figure
6-44. The combination of non-critical front current pair selection and 3-section like tuning
map allows fast and efficient calibration of the device.




                                       Page 45 of 144
                                                                        First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                  3ab4c127c833.doc




                                                                   23
                            Figure 6-44 – Wavelength tuning map.




                                                                             23
                    Figure 6-45 – An example of a rear-phase tuning plane.

Figure 6-46 shows SMSR and associated power variation (within 2.5 dB, with minimum
power of 13.5dBm) for a SMSR >40dB over 80 channels (50GHz spaced) of the ITU grid
(where channel 1 is at 191300GHz and channel 80 at 195950GHz), obtained at fixed SOA
(150mA) and gain currents (200mA), with no active levelling of the power. DS-DBR
produces over 40 mW ex-facet power in the free-running condition, so the integrated SOA
runs at a high input power and low gain (3-6 dB). To prevent injection of an amplified return
wave into the DS-DBR, which might disrupt the tuning characteristics, the structure
includes an angled output facet.




                                       Page 46 of 144
                                                                      First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                3ab4c127c833.doc




                                                                                            23
        Figure 6-46 – Side mode suppression ratio (left) and fibre coupled power (right).

Figure 6-47 shows RIN (typically ~150 dB/Hz) and linewidth (intrinsic linewidth < 1MHz)
measured for each of the 80 channels, resulting in transmission performance comparable
to high-quality DFB sources for external modulation.




                                                                                                  23
Figure 6-47 – Linewidth for each channel (red crosses) and RIN (blue squares) for each channel.

One of the typical integration steps is, as seen for other products, the addition of a co-
packaged Mach-Zehnder modulator (the MZ structure ensures wavelength-insensitive
operation across the C-band), that has been demonstrated with performances suitable for
10 Gb/s transmission24.
The back-to-back sensitivity for 10 Gb/s transmission over 100km of SMF28 fiber was
measured to be –21.5dBm for 10-9 BER (within expectation for a zero-chirp modulator, as
shown in Figure 6-48), and the power penalty for 100km transmission was < 3dB across
the band. No adjustment of MZM bias point or drive amplitude was made over the band.




                                        Page 47 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




   Figure 6-48 – Measured sensitivity across the C-band for 10Gbps transmission over 100km of
                                                                           24
                   SMF28. Insets show back-back eye and eye after 100km.

According to public information the DS-DBR tunable laser able to cover the C band or the L
band should be commercially available in the fourth quarter of 2004.



NTT (SSG-DBR)
NTT studied an SSG-DBR laser monolithically integrated with a semiconductor optical
amplifier (SOA), with tuning range is over 40 nm, 10 mW output power and with a built-in
wavelength locker and control circuit.
The SSG-DBR laser requires a complicated tuning control to obtain an arbitrary
wavelength. To improve the wavelength stability and simplify the wavelength control,
proper wavelength locker and control circuits have been studied25.
In order to monitor wavelength stability with the monitored output power, an evaluation of
the characteristics as a function of the driving current in the front and rear sections (Figure
6-49) have been studied (considering carrier-induced absorption losses in the passive
SSG-DBR sections, with a larger effect in the front SSG-DBR section, since the output light
has to pass through it).




                                       Page 48 of 144
                                                                      First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                               3ab4c127c833.doc




                                                                 25
                               Figure 6-49 – Output power map.

In the proposed stabilization circuit (Figure 6-50) a small sinusoidal reference signal was
added to the front SSG-DBR current (If) and a reference signal, which phase was shifted
by 90 degrees, is added to the rear SSG current (Ir). Therefore, each tuning current is
modulated like drawing a circle in the If-Ir plane. Error signals for If and Ir are generated by
the phase-sensitive detection with a lock-in amplifier, and fed back to the corresponding
SSG-DBR current. In this way, the reflection peaks of the two SSG-DBRs are adjusted to
the longitudinal mode.
The wavelength of the oscillation mode is controlled by a reference optical filter, which
generates an error signal that is fed back to the phase-control current.
The proposed circuit works well even for a change of +/-5°C and the wavelength can be
fixed even if the peaks of the two SSG-DBRs and/or the longitudinal mode are shifted
within +/-0.5 nm due to the degradation by aging.




                                        Page 49 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




   Figure 6-50 – Schematic of a simplified wavelength stabilization circuit. A and B are feedback
  currents to the front and rear SSG-DBRs for stabilizing the oscillation mode, and C is feedback
                                                                                   25
                  current to the phase control section for locking the wavelength.

In SSG-DBR lasers, the oscillation wavelength is rapidly switched by changing the tuning
current However, wavelength drift due to the thermal effect is observed; tuning-current
injection inevitably changes the temperature distribution in the laser and causes the
wavelength to drift after switching. The time constant of the thermal effect is of the order of
several µs to ms, which is very slow compared to the carrier induce refractive index
change. Therefore, it is necessary to compensate the thermal effect in high-speed
wavelength switching. Moreover, it is important to quickly stabilize the oscillation
wavelength after the switching.

ADC/Altitune (GCSR)
GCSR was originally patented by Swedish startup Altitun. It was bought by ADC
Telecommunications Inc that later closed the optical component business.
GCSR frequency stability has been studied26. As the laser ages, the tuning efficiency
decreases and the same operation point, i.e. the same currents, yields a lower frequency
than at the beginning of life.
Samples have been measured at different temperature and the largest resulting frequency
drifts were –3.6 GHz (50 °C) and –7.2 GHz (70°C), without stabilisation.
Testing in accelerated aging GCSR samples with a frequency and mode stabilization
algorithm, the same lasers showed no mode hops, negligible frequency drift (Figure 6-51)
and small variations of the SMSR.




                                         Page 50 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                               3ab4c127c833.doc




 Figure 6-51 – Frequency drift as a function of normalized aging time of three GCSR lasers (5 OPs
 per laser) with and without stabilization. The depicted aging time is estimated to correspond to at
                                 least 20 years in normal operation.


SYNTUNE (MG-Y modulated grating Y-branch laser)
The modulated grating Y-branch (MG-Y) laser is a monolithic electronically tuned
distributed Bragg reflector laser employing modulated grating reflectors with multiple
reflectivity peaks to achieve wide tuning27.
The MG-Y (Figure 7-10) is similar to the SG-DBR in that it uses the Vernier effect to
achieve wide tuning with two multipeak reflectors and has also similarities with GCSR
laser, in the sense that it has all tuning sections on the same side of the gain section, so
the light can exit the cavity without absorption, enabling higher output power across the
tuning range.
Thanks to its structure, MG-Y laser overcome some limitations of SG-DBR lasers (output
power varies significantly with tuning since the output light has to pass through the front
reflector) and GCSR lasers (long chip which reduces the cavity mode spacing and yields
somewhat lower side-mode suppression ratio).
The light is split by the use of a multi-mode interferometer (MMI), with S-bends to increase
the separation between the waveguides: each arm ends with a multi-peak reflector. In one
arm there is a differential phase section that can be used to adjust the phase difference
between the reflections. A common phase section is used to align the cavity mode with the
reflector peaks. The differential phase section has been evaluated as unnecessary, so 4
sections control will be possible (as for SG-DBR and GCSR lasers).




                                         Page 51 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                3ab4c127c833.doc




                                                                                          27
        Figure 6-52 – Schematic lay-out of the modulated grating Y-branch (MG-Y) laser.

Figure 6-53 shows a map of the output frequency as a function of the tuning currents in
both reflectors (tuning currents are low, less than 16 mA, which is important for fast
switching of the lasers since this results in reduced thermal transients when changing
channel).
The lasers can be tuned from 191.05 THz to 196.80 THz with side-mode suppression
ratios of more than 40 dB and an average output power of about 14 dBm, with variations of
1.2 dB (Figure 6-54).




 Figure 6-53 – Measured map of output frequency as a function of left and right reflector currents.
                                                                                            27
     Gain current: 150 mA, common phase current: 0 mA, differential phase current: 0 mA.




                                         Page 52 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                3ab4c127c833.doc




  Figure 6-54 – Side-mode suppression ratio (left) and front facet output power at a constant gain
                                                                  27
                              section current of 150 mA (right).

The modulated grating Y-branch laser is a InP/InGaAsP buried heterostructure device
composed by a MQW section that is butt-joined to the passive sections made in a InGaAsP
layer.
Manufacturing process is essentially the same as that for a standard DBR laser, with five
MOVPE steps. Stitching error free gratings are fabricated in the two reflectors by electron
beam lithography and wet chemical etching.
The first batch of MG-Y lasers was manufactured within the EU-funded project IST-2000-
2844 NEWTON. The project also develop concepts for a Ring Resonator Sampled Grating
laser diode based on the combined filter operation of a ring, giving a periodic transmission
spectrum and of a sampled grating, giving a multi-peaked reflection spectrum.
The Syntune S1500, based on Syntune's patented modulated grating Y-branch (MG–Y)
laser design, is packaged into a hermetically sealed 14–pin butterfly package, with an
integrated optical isolator and a wavelength locker.
S1500 preliminary characteristics are reported below28.

   Parameter         Symbol        Unit                             S1500
Tuning                  BT         nm      40
Bandwidth
Channel Spacing                  GHz     -
Tuning Time             TT         ms      - (50 ns as inherent limit)
Maximum Output          Po         mW      2 to 10
Power
Optical power           Pv          dB     1.5
variation
Wavelength              Wa         GHz     -
Stability
Maximum                 Fw        MHz      -




                                          Page 53 of 144
                                                              First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband              ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                         3ab4c127c833.doc




Spectral Width
Side Mode            SMSR        dB      40
Suppression
Ratio
Relative Intensity    RIN       dB/Hz    -
Noise
Polarization           Pe        dB      -
Extinction Ratio
Optical Isolation      Oi        dB      -
Maximum                Pe        W       -
Dissipated
Power



INTUNE (fast switching laser)
Intune proposes AltoNet1200TM Fast Wavelength-Switched Tunable Laser Module, with
characteristics reported in the table below29.

   Parameter         Symbol     Unit                     AltoNet1200TM
Tuning                 BT        nm      C-band
Bandwidth
Channel Spacing               GHz      50
Tuning Time            TT       nsec     200 (nsec)
Maximum Output         Po        mW      10
Power
Optical power          Pv        dB      +/- 1
variation
Wavelength            Wa        GHz      +/- 5
Stability
Maximum                Fw       MHz      15
Spectral Width
Side Mode            SMSR        dB      30
Suppression
Ratio
Relative Intensity    RIN       dB/Hz    -135
Noise
Polarization           Pe        dB      18
Extinction Ratio
Optical Isolation      Oi        dB      30
Maximum                Pe        W       -



                                        Page 54 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




Dissipated
Power



External Cavity Tunable Lasers
Tunable External Cavity Lasers (ECLs) are edge-emitting lasers containing a gain chip
(conventional FP laser chip) and separate gratings or mirrors to reflect light back into the
laser to form a cavity.
To tune the laser output a grating or another type of narrow-band tunable mirror is adjusted
in a way that produces the desired wavelength. This type of tuning (MEMS is a common
solution for this device) usually involves physically moving the grating or the mirror.
ECLs can achieve wide tuning ranges (greater than 40 nm), although the tuning speed is
determined by the mirror tuning (it can take tens of milliseconds to change wavelengths in
case of mechanical solutions). External cavity lasers are widely used in optical test and
measurement equipment due to the high purity of their emission together with very high
output powers (at least 20 mW) over a broad range of wavelengths.
Often, the major disadvantages of this kind of tunable laser are due to the presence of
mechanical movement witch gives reliability and qualification problems.
External cavity lasers exploiting feedbacks from a dispersive element such as a grating
(providing strong frequency selective feedback) are generally based on either a Littrow
cavity or Littman-Metcalf cavity design (Figure 6-55). In both designs, one facet of the gain
chip is antireflection coated and the light output is directed through a collimating lens into
the cavity on one side of the chip.
In the Littrow configuration a diffraction grating sends the beam back toward the active
section, while undiffracted beam serves as the output: tuning is achieved mechanically
rotating and translating the grating, which changes its effective pitch.
In the Littman-Metcalf configuration the diffraction grating diffracts the light to the mirror,
which reflects the beam back to the grating and the active section: tuning is achieved by
rotating or translating the mirror, which varies the effective cavity length.
Littman-Metcalf design has some advantages, such as cavity tuning without laterally
shifting the output beam, without the inclusion of an additional beamsplitter. Moreover,
because of the grazing incidence, a larger number of grating lines is illuminated, giving a
better frequency selection.
Besides these configurations, other solutions to change cavity length (or refractive index)
are possible.




                                       Page 55 of 144
                                                                        First Guideline for Technologies
                                  IST IP NOBEL "Next generation
                                  Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                       European Leadership"                              3ab4c127c833.doc




                                 laser




                         laser




Figure 6-55 – External cavity lasers: Littrow configuration (top) and Littman-Metcalf design (bottom)
                                The laser output is collimated by lens.




IOLON (ECL)
Iolon proposes a MEMS-tunable ECL, with a MEMS technology similar to the one
described for tunable filters in xxx, which has been Telcordia qualified and demonstrated in
tuning range of 5 THz in the C or L band (broader tunability is possible, at least in principle,
because the bandwidth of the gain chip is greater than 10 THz)30.
Figure 6-56 shows the structure of the device, with the mirror movements required for
tuning. The depicted solution overcomes typical limitations in mode-hop-free tuning ranges
due to tradeoffs in the MEMS device between angular travel, mirror translation and
vibrational sensitivity. The concept is to select a particular path (multiple optical paths exist
for different frequencies in the same resonator) by tilting the mirror and adjusting the out-of-
plane collimating lens position so the desired path returns to the gain medium, without
requiring a change in the actuator angular travel. This allows to enlarge both the tuning
range and the mode-hop-free tuning range, exploiting all the tunability band enabled by the
gain chip.




                                          Page 56 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                               3ab4c127c833.doc




  Figure 6-56 – Schematic showing the multiple paths in the laser resonator, illustrating how lens
                                                                                          30
          position and mirror tilt can be used to select a particular path and frequency.

It can be shown that if the mirror is tilted while leaving its rotation axis parallel to the grating
groove direction, then the lens alignment remains constant over a tuning range of several
degrees, so incorporating a suitable actuator makes it feasible to actively switch between
the different paths as part of the tuning process.
Figure 6-57 shows a view of the component.




Figure 6-57 – SEM view of the actual components. Note that the lens MEMS actuator is behind the
                                                               30
                                   grating and is not visible.

The inset in Figure 6-58 shows the fiber coupled output power measured at the mirror
center position as the lens MEMS voltage is varied in a resonator with a 0.5 degree mirror
tilt. The two strongest peaks at 54 and 68V exhibit the spectra shown. The extinction ratio
between the two paths can be made larger by increasing the mirror tilt and values up to 1.5
degrees have been used.
SMSR is dominated by the modes adjacent to the main mode, not oscillation at the
alternate resonator path, and is similar for both C and L band.




                                         Page 57 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




Figure 6-58 – Spectra showing the two frequencies that can be obtained at the same mirror position
                                                                30
                                  by varying the lens position.

Figure 6-59 shows the laser threshold and the average power observed with a 180 mA
laser current for the two lens positions. The power output and slope efficiency for the L-
band path is higher than for the C band path, even at the same frequency. Channel
switching times of less than 10 ms could be obtained within a single band.




   Figure 6-59 – Laser threshold and average fiber coupled output power for the laser in the two
                                                                                     30
            configurations. The output power is greater than 10 dBm over both bands.

Since mirror translation required to achieve continuous tunability in the C band
configuration is less than for the L band, the mirror zero mode-hop pivot point is closer in
the C band than it is in the L band, to the point where the beams intercept the grating.
Properly positioning the mirror actuator, the few remaining mode hops could be eliminated
during a scan using an active adjustment of the cavity length using a piezoelectric device
attached to the grating, and it is possible to tune over the full C band tuning range without a
mode hop.




                                         Page 58 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




Active cavity length control makes it possible to maintain laser frequency through
temperature excursions, mechanical shocks and device aging, avoiding mode hops.
Due to asymmetry of the tuning curve, ECL performances (measured by the side mode
suppression ratio), is optimized for non-zero detuning, thus a standard dithering is
unnecessary and simple proportional control can be used to control the detuning for
optimum SMSR31.
Figure 6-60 shows the wavelength control scheme, including a wavelength locker (WLL)
with four photodiodes, a linear dielectric filter and an etalon. PZT changes the round trip
cavity length by 3-5 wavelengths, and the silicon virtual-pivot MEMS actuator scans a
combination of filter center frequency and cavity length to allow a tuning range of 5 THz.
Once the laser is locked to channel, the laser has good long-term stability and provides
optical powers exceeding 20 mW over a 5 THz tuning range with a linewidth (55 dB)
suitable for use in 2.5, 10 or 40 Gbps externally modulated systems.
Stability of ± 1.0 GHz and ± 0.2 dB over 5 THz tuning range have been demonstrated with
case temperatures between -10 and 70 ºC.
The ECL is assembled on a ceramic substrate and bonded to a thermoelectric cooler
(TEC) for thermal stabilization. The gain chip is a InGaAsP/InP multiple-quantum-well.




 Figure 6-60 – Schematic of the ECL and WLL showing the components used to tune and lock the
                                               31
                                           ECL.

Iolon proposes the Apollo™ tunable laser module, operating in C or L band on 200
chanels at 25 GHz spacing, with characteristics reported in the table below32.

   Parameter        Symbol       Unit                           Apollo
Tuning                 BT        nm      40
Bandwidth
Channel Spacing                GHz     25 (or 50)
Tuning Time            TT         ms     100
Maximum Output         Po        mW      20
Power




                                        Page 59 of 144
                                                                    First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                            3ab4c127c833.doc




Optical power            Pv         dB     +/- 0.25
variation
Wavelength               Wa        GHz     +/- 1
Stability
Maximum                  Fw        MHz     2 (self-heterodyne 3.5 s delay time)
Spectral Width
Side Mode              SMSR         dB     > 43
Suppression
Ratio
Relative Intensity      RIN       dB/Hz    < -145 (1 MHz-10 GHz)
Noise
Polarization             Pe         dB     20
Extinction Ratio
Optical Isolation        Oi         dB     45 (typ.)
Maximum                  Pe         W      5.5
Dissipated
Power



INTEL (ECL)
Intel proposes a full C-band tunable ECL (TTX11500)33, with a scheme depicted in Figure
6-61, that uses an external cavity design with InGaAsP/InP gain medium and intracavity
etalon based, thermally actuated, widely tunable filters fabricated from silicon wafers.




                                                                                      33
             Figure 6-61 – Intel® TTX11500 Full C-Band Tunable Laser Block Diagram.

The ECL (with no moving parts) is wholly mounted on a ceramic platform placed on a
thermo-electric cooler (TEC) within a 14-pin butterfly sized package, including output optics
(collimating lens, optical isolator, beamsplitter and monitor photodiode, fiber coupling lens,
and a pigtail that couple the laser emission into Polarization Maintaining fiber).




                                          Page 60 of 144
                                                                          First Guideline for Technologies
                                 IST IP NOBEL "Next generation
                                 Optical network for Broadband                      ee787c6d-c81c-433e-a3ea-
                                      European Leadership"                                 3ab4c127c833.doc




The tuning filters are two silicon etalon filters with slightly different periods, in order to
exploit Vernier effect for wider tunability (Figure 6-62): filters are thermally actuated, taking
advantage of silicon’s high thermo-optic coefficient. Any wavelength in the C-band can be
addressed with a small temperature adjustment of both individual etalons, spectrally
translating transmission maximum, within 1 second.




 Figure 6-62 – Transmission spectra of the tuning filters with lasing on a mode where the spectral
 peaks align (top). Net transmission through the composite filter (lower left). Typical resultant laser
                                                              33
                                      spectrum (lower right).

Separation between gain and tuning sections allows turning off the gain medium while
tuning, preventing laser emission at any wavelength other than the final target wavelength.
The power control implementation uses a partial reflector that splits a fraction of the output
light onto a monitor photodiode, so that a digital control loop adjusts the injection current to
achieve the calibrated photocurrent target.
Figure 6-63 and Figure 6-64 show start of life performances of a tunable laser prototype,
obtained tuning over the C band in 50 GHz steps with 8 different case temperatures,
compared with common target specifications (red lines: these limits does not correspond to
product target specifications). As can be seen, measurements are well within limits for
output power (13 dBm with +/- 0.5 dB), optical frequency deviation from ITU-T grid (+/- 2.5
GHz), Side Mode Suppression Ratio (40 dB) and tuning time (25 seconds, considered as
“cold start” tuning time).
The tunable laser has no moving parts, which helps achieve operational shock and
vibration resistance.




                                           Page 61 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




Figure 6-63 – Output power measured (in 50 GHz steps over the C-band and repeated at 8 different
    case temperatures) on an Intel® C-band tunable laser prototype. The nominal output power
                                                                                  33
               specification is 13dBm, wiht maximum allowed deviation of ±0.5 dB.




                                       Page 62 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




   Figure 6-64 – Start of life wavelength accuracy (top), SMSR (middle) and tuning time (bottom)
                                                                                          33
       measured tuning across C-band in 50 GHz steps at 8 different case temperatures.

Efficient, high-yield manufacturing processes developed by Intel for 10 Gb/s transmitters
have been extended for the tunable laser application, in order to guarantee both broad
tunability and excellent wavelength accuracy in closer channel spacing over product life.




                                         Page 63 of 144
                                                                        First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                3ab4c127c833.doc




Moreover, fast calibration methods have been developed and knowledge-based reliability
assessment has been conducted to demonstrate the robustness of the design34.
The ECL (Figure 6-65) is bounded by the front facet of the gain chip at the front and the
lithium niobate mirror at the back. The undesired reflectivity of the interior facet of the gain
chip is reduced by conventional anti-reflection coatings and by a bent waveguide path.
The module is assembled on a flat ceramic substrate upon which is deposited a patterned
metallic layer. Optical chips (laser, photodiodes) and electronic components are mounted
with an accuracy of approximately 10 microns by pick and place automation using machine
vision. Optical components such as lenses and fibers are mounted on etched and formed
flexible metallic elements (flexures).
The module optical alignment consists of four active alignments with the flexure technology
(two collimating lenses, one focusing lens, and one optical fiber) and two active alignment
components that require special thermal conductivity management, which are epoxied
down (back cavity mirror and tunable filter subassembly).




    Figure 6-65 – Solid model of the temperature-tuned external cavity laser assembled on the
quasiplanar laser welded manufacturing platform. Visible are welded flexures supporting the optical
  fiber, collimating lenses and a prism that sends a small fraction of the output beam power to a
 monitor photodiode. At upper right are the tunable filters and the partially obscured lithium niobate
                                                         34
                                            end mirror.

The tunable laser separates wavelength selection and tuning from the gain medium, in
addition to the wavelength reference and power monitor and implements them by hybrid
assembly techniques, avoiding loss by replacing absorptive regions of the InP used for
wavelength selection in DBR-like implementations with equivalent implementations in low
loss materials.
The intracavity tunable filter serves also as the wavelength reference. Wavelength locking
(cavity length locking) is achieved using the electro-optical dithering of the lithium niobate
substrate for the ECL cavity mirror.
Thermally tunable filters are micromachined silicon etalons (Figure 6-66). Microfabrication
is used to create thermal isolation by a silicon nitride suspension, and for electrical



                                          Page 64 of 144
                                                                       First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                         ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                    3ab4c127c833.doc




connectivity for local heating and temperature monitoring (the integrated Resistive Thermal
Device –RTD- thermometer uses a platinum sensing element).
Temperature monitoring is critical since it is correlated to wavelength accuracy, and it is
improved by the thermal conductivity of the silicon etalon that suppresses temperature
gradients. A four-wire probe is created by microfabrication to isolate the RTD function from
stray series resistance from wire-bonding and other sources. The temperature control
achieved by these measures supports, in conjunction with factory calibration, the role of the
wavelength filter as a stable and accurate wavelength reference.




                                                                               34
                     Figure 6-66 – Details of the Si thermally tuned filter.

A knowledge-based reliability assessment approach has been chosen to evaluate
product robustness to environmental stresses. Multiple test vehicles have been used to
isolate failure mechanisms associated with functional areas.
      High temperature and high-biased stress tests on laser chip on submount to assess
       the degradation mechanisms (power, optical spectrum) and the failure rate.
      Tests on front optics test vehicle to evaluate the robustness of the coupling
       efficiency between the laser output and the fiber output and the monitor photodiode
       responsivity.
      Tests on external cavity test vehicle to evaluate the robustness and mechanical
       stability of the external cavity end mirror and the collimating lens stability.
      Tests on tuner subassembly test vehicle to verify the robustness of the two
       thermally tuned etalons (resistance of the thermal detector, the optical transmission,
       and the angle tuning of the etalon).
Specific tests were focused on failure modes associated with the micromachined thermally
tuned silicon etalons. Highly accelerated testing has been conducted to analyse the time to
failure and the acceleration factors. The observed frequency drift of tuners during a 100-
hour test at 140°C and a 2000-hour test at 85°C was below 200 MHz.



                                        Page 65 of 144
                                                                        First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                3ab4c127c833.doc




Based on stress test and evaluation of acceleration factor with temperature, the median
time to failure for nominal conditions (Figure 6-67) has been computed (beyond 130 years).
The results of prequalification on complete product demonstrated the robustness of the
design (for example, the frequency stability during the mechanical stress tests was max
300 MHz of frequency shift out of a 2.5 GHz specification).




    Figure 6-67 – Prediction of the time to failure as a function of temperature. The three curves
                                                                                  34
                   correspond to the predictions for 1%, 10%, and 50% failure.

Intel TTX11500 device characteristics are reported in the table below35.

   Parameter          Symbol       Unit                            TTX11500
Tuning                   BT         nm      35 (from 1528.77 nm to 1563.86 nm)
Bandwidth
Channel Spacing                  GHz      25
Tuning Time              TT         ms      - (few seconds?)
Maximum Output           Po         mW      20
Power
Optical power            Pv         dB      -
variation
Wavelength               Wa        GHz      +/- 2.5 for 50 GHz spacing
Stability
Maximum                  Fw        MHz      -
Spectral Width
Side Mode              SMSR         dB      > 45
Suppression
Ratio
Relative Intensity      RIN       dB/Hz     < -145
Noise



                                          Page 66 of 144
                                                                  First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                           3ab4c127c833.doc




Polarization           Pe        dB      -
Extinction Ratio
Optical Isolation      Oi        dB      -
Maximum                Pe         W      2.5 (typ.)
Dissipated
Power



PRINCETON OPTRONICS (ECL)
Princeton Optronics proposes Powersweep™ 2000, a tunable laser module based on
external cavity design. The wavelength selection is performed via movement of a piezo-
electrically controlled etalon, achieving continuous tuning of the laser, which can be used
for any ITU-T grid frequency from 100 to 12.5GHz of channel spacing (currently 25 GHz
channel spacing is supported). Other characteristics of the product are high power (up to
50mW), +/- 1GHz of frequency accuracy and low power dissipation (3W).
Powersweep™ 2000 tunable laser characteristics are reported in the table below36.

   Parameter         Symbol      Unit                    Powersweep™ 2000
Tuning                 BT        nm      35 (from 1528.77 nm to 1563.86 nm)
Bandwidth
Channel Spacing               GHz      25
Tuning Time            TT        ms      500
Maximum Output         Po        mW      20
Power
Optical power          Pv        dB      +/- 0.5
variation
Wavelength             Wa       GHz      +/- 4.5 (EOL, over temperature operating range)
Stability
Maximum                Fw       MHz      0.15
Spectral Width
Side Mode            SMSR        dB      50
Suppression
Ratio
Relative Intensity    RIN       dB/Hz    < -145
Noise
Polarization           Pe        dB      20
Extinction Ratio
Optical Isolation      Oi        dB      25
Maximum                Pe         W      5
Dissipated
Power



                                        Page 67 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                3ab4c127c833.doc




ALCATEL/(AVANEX) (ECL)
A concept for an external cavity laser including sampled fiber Bragg grating and able to
tune over 12-channel at 200GHz steps has been presented37, basing on studies
demonstrating that a weak current variation applied to a single electrode induces a Vernier
step tuning of 15nm between a Sampled Fiber Bragg Grating (SFBG) and a two sections
laser diode38.
The cavity length is adjusted to be slightly different from the reflectivity peak spacing.
Therefore, a Vernier effect is possible between the comb-like Fabry-Perot modes and the
reflector peaks. When one Fabry-Perot resonance coincides with a peak of the reflector,
lasing occurs at this frequency. A current injection in the phase section allows a shift of the
Fabry-Perot comb and induces a mode jump of the laser to the next coincidence. The
tunability of such a laser is mainly limited by the mismatch between the Laser Diode and
the Sampled Fibre Bragg Grating Free Spectral Range.
The structure allows selecting the wavelength with only one control (current) and, due to
low current variations, the output power remains constant.




  Figure 6-68 – Schematic of the structure and principle of the Vernier effect between Fabry-Perot
                                                                            37
                     cavity and Sampled Fiber Grating reflectivity peaks

Over 20nm 200Ghz step tuning has been obtained by varying the phase current from 0.5 to
8.5mA. Over this range, the SMSR is in excess of 40dB and the power variation is less
than 0.6dB around 8dBm at 45°C.
A fast switching time between channels of less than 20ns has been measured. The design
is potentially compatible with a full C-Band 50GHz-spacing tunablility.




                                         Page 68 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




PIRELLI (ECL)
Pirelli proposes DTL-C13, a C-band tunable ECL, based on voltage-controlled mirror with
no moving parts.
In the following table DTL-C13 characteristics are reported39.

   Parameter         Symbol      Unit                            DTL-C13
Tuning                 BT         nm     35
Bandwidth
Channel Spacing                GHz     50 (25 GHZ spacing compatible)
Tuning Time            TT         ms     10
Maximum Output         Po        mW      20 (max. -20 turn off)
Power
Optical power          Pv         dB     +/- 0.5
variation
Wavelength             Wa        GHz     +/- 1.5
Stability
Maximum                Fw        MHz     0.5 (ist.)
Spectral Width
Side Mode            SMSR         dB     50
Suppression
Ratio
Relative Intensity     RIN      dB/Hz    -145
Noise
Polarization           Pe         dB     25
Extinction Ratio
Optical Isolation      Oi         dB     25
Maximum                Pe         W      5
Dissipated
Power

VCSEL Tunable Lasers
In VCSELs the resonant cavity is perpendicular to the semiconductor layers and is formed
by mirrors (alternating layers of semiconductor material with different refractive indexes to
form Bragg reflectors) located above and below the active section.
Tunable VCSELs are made with a movable top mirror (implemented using MEMS
technology) allowing tuning changing the cavity length. Vertical-cavity lasers are potentially
tunable over a wide range, on the order of 40 nm and the switching time is typical of MEMS
(order of milliseconds). Power output is usually only a few milliwatts, although work with
larger cavities and emitting apertures promises to increase power levels. Moreover
VCSELs at 1300 and 1550 nm have manufacturing and material constraints.




                                        Page 69 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




Since fabricating VCSELs requires only a single process growth phase, manufacturing
them is much simpler than producing edge emitters. VCSEL manufacturers can also exploit
wafer-stage testing, thus eliminating defective devices early in the manufacturing process,
saving time, and improving overall component manufacturing yields.
Tunable VCSEL are normally electrically pumped (monolithic, wide tunability, low power
levels, direct modulation is possible, generally suitable for metro-access applications).
Optically pumped reach greater output power (1–10 mW), but require complicated hybrid
packaging and main developments have been discontinued.


Tunable Lasers based on other technologies
In the following a short summary of some studies, among several research activities
investigating novel methods to achieve wide tunability with stable optical characteristics, is
reported.
A structure based on InP with photonic crystal mirrors couplers and a monolithically
integrated amplifier section for stabilized power output has been demonstrated40. The laser
consists of three longitudinally coupled ridge waveguide cavities that are coupled through
photonic crystal mirror segments. A laterally defined binary superimposed grating provides
complex-coupled distributed feedback into two of the laser cavities. The wavelength can be
switched with current controls over a range of 26 nm. Through adjustment of the current
into a third amplifier section, the output power can be stabilized over the tuning range of
the laser.
Researchers from Fujitsu have demonstrated41 a tunable lasers combining
semiconductor optical amplifiers and acousto-optic tunable filter (AOTF), achieving a
90 nm tuning range with uniform power and stable single-mode oscillation. AOTF has a
wide tuning range and fast switching speed (several microseconds). Additionally, an AOTF
provides stable operation against shock and vibration due to its non-mechanical structure.
Researchers from NTT have demonstrated42,43 a tunable laser consisting of a ladder
filter, a ring resonator, and an SOA. The ladder filter selects one channel from the
periodic channels of the ring resonator, therefore controlling the lasing wavelength with
only the tuning current of the ladder. The device, fabricated using InP-InGaAsP materials,
exhibits 31-channel 100-GHz-spacing.
Researchers at UCL have demonstrated44 a tunable source based on optical injection
locking (OIL), achieving <12 ns switching time between 30 channels over a tuning range
of 4.5 nm (18 GHz channel spacing) and <1kHz locking error.
Researchers at NEC have demonstrated45 a tunable laser consisting of a silica waveguide
double-ring resonator and a semiconductor optical amplifier, achieving more than 45-
nm wavelength tuning and +4.7 dBm fiber coupled power.



Tunable Lasers comparison
Tunable lasers can be evaluated with respect to target performances.
In terms of optical performances, ECLs are probably a good choice, considering models
without moving parts; major limitation is the tuning speed, which is at the best in the range
of milliseconds.




                                       Page 70 of 144
                                                                   First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                             3ab4c127c833.doc




In case faster tuning is needed, multisection DBRs are good candidates (even if current
products normally have tuning speeds in the milliseconds range, due to control electronic),
that anyway provids good characteristics and intrinsic integrability with modulators.
VCSELs are still not mature, low performances and more suitable for metro applications.
In the following table, some key points are reported to compare widely tunable lasers
(please also refer to Nobel D3 - Table 3-1: Comparison of tunable laser technology).

  Tunable laser                Advantages                           Disadvantages
DFB array             Well suited for integration, with   Optical performances/tuning range
                      estabilished manufacturing          tradeoff
                      process.
                                                          Slow tuning time (thermal tuning -
                      Direct modulation possible          seconds)
Multisection DBR      Inherently fast switching           Evolving manufacturing methods,
                      speed                               with large chip area and
                                                          consequently yield decrease
                      Direct modulation possible
                                                          Complex control
                      Modulator integration
ECL                   High spectral purity and output     Medium tuning time (milliseconds –
                      power                               hundreds of milliseconds)
VCSEL                 Low cost technology                 Low performances (optical output
                                                          power)


Tunable laser reliability is of course a key aspect to be considered: Telcordia
Technologies Inc. standard GR-468-CORE is the reference specification for fixed-
wavelength lasers reliability. Updates to this standard should be needed to consider
fundamental aspects of tunable lasers reliability, in order to provide a common reference
for vendor tests.
Despite announcements of Telcordia qualification of MEMS-based products, still this
technology has some open reliability and wavelength stability issues. So, for instance, even
if DFB lasers are extremely reliable, products based on DFB array with MEMS selection
should be carefully analysed.
How the tunable laser controls wavelength selection/stability and how ageing affects this
process is also a major point and encompasses both the wavelength tuning (correct
selection of the channel) and the long-term stability (to maintain the selected channel over
device lifetime).
Main degradation causes, that have a different impact for different technologies, range
from modifications of cavity length (physical movements, MEMS charging), to changes in
refractive index (due to changes in temperature and carrier lifetime). Degradation in DBR
grating has been limited due to technological improvements.




                                       Page 71 of 144
                                                                     First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                             3ab4c127c833.doc




References



  7       ROADM – Reconfigurable Optical Add-Drop
          Multiplexers
 Reconfigurable OADMs (ROADMs) are key building blocks of all-optical networks, since
 they can allow the dynamic rerouting of selected wavelengths under the supervision of the
 optical network control plane.
 An introduction to ROADMs is discussed in Nobel deliverable D3 chapter 3.2.3, where the
 main characteristics are reported. Before discussing the aspects on which this section will
 focus on, it is worth to recall some key points that are driving ROADM technical
 requirements.
 With respect to fixed OADMs, ROADMs provide inventory savings (a single widely tunable
 ROADM unit can replace several fixed OADM components corresponding to different A/D
 lambda subsets) and planning benefits (uncertainty of traffic matrix evolution that leads to
 stranded lambdas when adopting fixed OADMs). Under this point of view, a ROADM does
 not involve additional requirements compared with a fixed OADM (switching time is not an
 issue). The only mandatory requirement is that ROADM tuning must be hitless (service
 interruption only on the signal of the channel being reconfigured without any disruption on
 other channels).
 In a dynamic network one of the key functions is the possibility to perform a lambda
 rerouting to/from any client interface under a software control: this necessarily requires a
 not blocking add/drop (here and in the following, the term ‘not blocking A/D’ refers to a
 ROADM able to route any lambda to any port). The required switching time should be
 (provisional requirement derived from a possible switching time allocation to ROADM
 subsystem, to be refined from requirements coming from other WPs) around few tens of
 milliseconds.
 Other features that are highly desirable in a dynamic network scenario are relevant to
 signals conditioning, i.e. the possibility to perform optical channel level equalization, to
 optical channel power level monitoring and ROADM cascadeability (limited disruption on
 the transiting channels to to self-filtering and dispersion accumulation).
 Moreover, when ROADMs are adopted in configurations where ASE can accumulate (e.g.
 rings), ROADM properties with respect to ASE filtering can also be considered.
 Another point that should be evaluated is relevant to the number of add/dropped channels.
 As a provisional requirement, that could be refined from the outcome of other Nobel WPs,
 we assume that two configurations cover the majority of network applications: ROADMs
 with 4/8 add/dropped channels and ROADMs with 100% add/drop capacity.
 A feature that can be provisionally considered is the branching function, i.e. the possibility
 to perform pure optical routing (a single lambda) and branching (a lambdas subset),
 requiring that ROADM is able to perform A/D of a defined lambda set.



                                        Page 72 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




From a structural point of view, ROADM term is normally applied to a subsystem able to
provide add/drop flexibility with respect to a fixed OADM; but beside this aspect, another
added functionality to be considered with respect to fixed OADM applied to a static point-
to-point optical link is the possibility to allow rerouting among different optical links, as
required in a mesh topology. The possibility to support applications with node degrees
higher than two is a possible requirement for ROADMs applied in a dynamic network.
Therefore ROADMs should support current network topology (point-to-point on long haul
and ring on metro core) and provide an evolutionary path to support dynamic network, with
mesh topology.
In the following paragraphs will be discusses:
   -     Main features and parameters definitions, in order to compare different
         technologies and architectures
   -     Main internal architectures for ROADMs
   -     ROADM technologies and existing/proposed products


 7.1      ROADM main characteristics

 7.1.1       Connectivity
ROADM characteristics in terms of connectivity can be summarized as the network-level
features describing the routing capability of the ROADM.
ROADM connectivity parameters are describe below:

   Parameter         Symbol      Unit                       Description
Operating               BO       nm     It is the maximum bandwidth allowed for the pass-
Bandwidth                               through traffic.
Tuning                  BT       nm     It is the maximum bandwidth that can be covered
Bandwidth                               with tuning on add/drop ports. It should be noted that
                                        the tuning bandwidth could be different from the
                                        actual bandwidth of the pass-through signal (e.g. a
                                        ROADM can let the full C-band transiting and A/D
                                        only few channels). Moreover the tuning bandwidth
                                        is intended as the combination of the tuning
                                        bandwidth of each add/drop port, so 2 add/drop
                                        ports able to tune on 16 nm each can result in a
                                        ROADM having a 16 nm tuning bandwidth (the two
                                        ports tuning on the same band) or 32 nm tuning
                                        bandwidth (the two ports tuning on different bands)
                                        or in any value in between (partial overlap between
                                        the two ports tuning range). For this reason the
                                        ROADM tuning bandwidth should be considered
                                        together with the Routing Power.
A/D ports                K        -     Number of add/drop ports
Channel Spacing               GHz     Supported spacing between add/dropped optical
                                        channels




                                        Page 73 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                              3ab4c127c833.doc




Tuning Time             TT       ms     It is the maximum time to change the add/drop
                                        wavelength
Routing Power          RP          -    Routing power is a number (ranging from 0 to 1)
                                        representing the flexibility of the ROADM (being
                                        equal to 1 in case of full flexibility that is, as
                                        described above, not-blocking A/D). Please refer to
                                        the text below for Routing power definition. As a
                                        simplified concept, in the document we will also refer
                                        to the ‘any-lambda-to-any-port’ as the capability of a
                                        ROADM to add/drop any possible wavelength within
                                        its tuning range at any possible add/drop port
Node Order             NOrd        -    Number of optical links supported. It is equal to 2 for
                                        standard OADM.


The Routing Power parameter (actually split into two parameters, Global Routing Power,
RPG, and Segmented Routing Power, RPSi) has been proposed46 in order to derive a high-
level indicator of the flexibility of the add/drop function.
The Global Routing Power is a function of the number of wavelengths in the WDM
spectrum (N) and the maximum number of wavelengths allowed for simultaneous add/drop
(K): The full add/drop (i.e. K=N) offers of course the most flexible routing. Nevertheless, N
and K do not fully describe ROADM routing capability, since possible constraints in the
specific wavelength to be add/dropped are not included (e.g. certain lambdas can be fixed
on the express path, or each port can have a different accessibility to the WDM comb, as
usual in band-type design). In order to evaluate the connection flexibility of the ROADM it is
necessary to compute the number of individual connection states to which it can be set.
A ROADM connection state is described as a vector (or with a matrix47) with the dimension
equal to the number of wavelengths N and with elements equal to 1 (if the corresponding
wavelength can be add/dropped) or 0 (no add/drop possibility).
         log(# of connection states supported by ROADM)
RP G 
                              Nlog(2)
In case of a full flexible ROADM (no constraints on add/drop) has RPG = 1, while RPG = 0
corresponds to a fixed OADM.
It should be noted that the Global Routing Power does not reflect the constraints on the
add/drop port selection, since it does not include the concept of add/drop ports and the
flexibility in routing the add/drop wavelength in any port; the Segmented Routing Power
represents this further concept. RPSi is defined as the RPG but the count of connection
states is limited to those accessible within the group of ‘i’ add/drop ports; when i=K (all the
add/drop ports are included in the segmentation group) the Segmented Routing Power
corresponds to the Global Routing Power.
A discussion of Routing Power applied to some ROADMs is reported in the paragraph
relevant to ROADM main architectures
In the assessment of existing or proposed technology, we will consider the following targets
(many of them being provisional values, to be modified according to the results of specific
Nobel activities in other WPs.




                                        Page 74 of 144
                                                                        First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                      ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                 3ab4c127c833.doc




   Parameter          Symbol     Unit                      Provisional requirements
Operating/Tuning        BT       nm         Full C-band.
Bandwidth
A/D ports               K         -         4 add/drop ports for basic ROADMs, 100% A/D for
                                            hub ROADMs
Channel Spacing               GHz         50/100 GHz
Tuning Time             TT       ms         <10-50 milliseconds
Routing Power          RP         -         RPG = 1
                                            RPS1,…N = 1
                                            (that means full flexibility: ‘any-lambda-to-any-port’)
Node Order             NOrd       -         Node Order equal to 2 (the possibility to support
                                            higher order nodes can be considered as an
                                            advantage for specific technologies)



 7.1.2       Optical Characteristics
In this paragraph are reported general features and parameters definition of the IN-OUT,
IN-DROP, ADD-OUT ports and filtering requirements. When considering the filtering effects
it should be intended as the effective filter seen by the considered channel, whenever it is
the Add, Drop and Pass Through channel.

         Parameter              Symbol          Unit                   Description

            IN-OUT
Pass Through Loss of non-          PT           dB    it is the insertion loss of the channels that
dropped channels (express                              are not dropped.
channels)
Pass Through Loss                  UPT           dB    it is the difference between the insertion
Uniformity of non-dropped                              loss of the best-case and worst-case
channels (express                                      express channels
channels)
Rejection of dropped                  CIB        dB    it is the difference between the intensity
channels                                               of an input channel to be dropped and
                                                       the residual intensity of the same channel
                                                       measured at the out port when it is
                                                       dropped. A good rejection of the dropped
                                                       channels is needed to avoid in-band
                                                       crosstalk between dropped channels and
                                                       added channels.

            IN-DROP
Drop Loss                             d         dB    it is the insertion loss of a dropped




                                            Page 75 of 144
                                                                     First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                             3ab4c127c833.doc




                                                     channel
Drop isolation of the              COBda       dB    it is the difference between the intensity
adjacent channels                                    of the dropped channels and the intensity
                                                     of the adjacent (non dropped) channels
                                                     measured ad the drop port (worst-case)


Drop isolation of non             COBdna       dB    it is the difference between the intensity
adjacent channels                                    of the dropped channels and the intensity
                                                     of the non adjacent (non dropped)
                                                     channels measured ad the drop port
                                                     (worst-case)

         ADD-OUT
Add Loss                            a         dB    it is the insertion loss of an added
                                                     channel
Add isolation of the adjacent      COBaa       dB    it is the difference between the intensity
channels                                             of the added channels and the intensity
                                                     of the adjacent (non added) channels
                                                     measured at the out port.
Add isolation of non-             COBana       dB    this is the same feature described for the
adjacent channels                                    adjacent channels.

        FILTERING
Filter Central Wavelength          prec        %    it is the allowable offset from the ITU grid,
Accuracy                                             required after the filter tuning.
Filter Pass-band                    BC         GHz   Spectral width (or fraction of the channel
                                                     spacing), measured at –1, -3, -20, -30 dB
Chromatic dispersion               GVD         ps/   it is the maximum amount of chromatic
                                                     dispersion induced by the filter.
                                               nm

Polarisation Dependent             PDL         dB    it is the difference between ILmax and ILmin
Loss                                                 experienced by the various polarization
                                                     states
Pass-band ripple                              dB    it is the difference between ILmax and ILmin
                                                     in the channel bandwidth for a given
                                                     added or dropped channel in any state of
                                                     polarization and temperature.


For this family of parameters, no provisional targets are provided, since different
combination of values can be accommodated depending on the specific network design.
Anyway, in the paragraph relevant to the assessment of ROADM technologies, values
different from typical ranges will be discussed. We assume that filtering characteristics
shall allow managing 10 Gb/s channels.




                                           Page 76 of 144
                                                                   First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                            3ab4c127c833.doc




 7.1.3        Architectural and Operational Characteristics
In this section some other features, linked to operational as well as structural aspects, are
discussed.

          Parameter                                     Description
Hitless                         R OADMs interrupt the service only on the signal of the
                                channel being reconfigured without any disruption on other
                                channels.
East-West separation            In order to support optical layer protection mechanisms
                                there should not be a single point of failure affecting both
                                the West and East sides
Optical channel equalization    In order to support optical path management in a dynamic
                                all-optical network there should be the way to manage the
                                optical power level of each transiting (or at least add/drop)
                                channel
Optical power monitoring        In order to provide information to performance monitoring
                                functions
Branching                       Ability to perform add/drop of a lambda set on each
                                add/drop port
Operating                       Power consumption, operating temperature range
Control                         Control interface and protocol
Scalability                     ROADM should be conceived in such a way to allow low
                                first-lambda cost, with pay-as-you-grow approach
Cascadeability                  Since reconfigureability is achieved through filtering (or
                                anyway through mechanisms resulting into filtering
                                effects), an optical signal traveling through several
                                ROADM will be affected by penalties due to self-filtering
                                and/or dispersion accumulation. Various aspects play a
                                role in this characteristic (filters shape, phase and center
                                lambda stability, as well as characteristics of the
                                transmitting laser). Another point to be considered in rings
                                is the prevention of ASE accumulation


With respect to the East/West separation, some evaluation can be made on the internal
architecture of the basic components of a ROADM, schematically represented in Figure
7-1: two different connectivity schemes can be envisaged between line ports and local
ports. These connectivity patterns should be traced back to the way in which elementary
modules (performing basic operations of input/express, add/output, input/drop) of the
ROADM subsystem can be grouped together. The choice of the connectivity pattern can be
important for reliability considerations.




                                       Page 77 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                  3ab4c127c833.doc




                 West line ports                                 East line ports




                                          Local ports

                    Figure 7-1– General operation for a ROADM subsystem.




                 West line ports                                  East line ports




                                          Local ports

             Figure 7-2 – A possible connectivity pattern for the ROADM subsystem.


The first connectivity pattern can be called bi-directional pattern and is shown in Figure 7-2.
It presents the following features:
          each optical tributary is connected in a bi-directional way to a line side only
           (West or East),
          in case of line breaks, (e.g., on the West side), the traffic on the opposite side of
           the node (say, East) is not affected,
          it reaches an automatic balance between transport distances on both directions
           (an important issue for long-distance applications),
          it is more inclined for application to non-protected traffic patterns,
          it is neutral for application to protected traffic patterns,
          it is suitable for point-to-point and bi-directional ring applications,




                                        Page 78 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                  3ab4c127c833.doc




           a “horizontal” grouping of basic modules (see Figure 7-4) does not guarantee
            any protection in case of fault of one of these modules; this fault is equivalent to
            a simultaneous line break on both West and East node sides;
           a “vertical” grouping of basic modules (see Figure 7-5) does guarantee
            protection in case of fault of one of these modules; this fault is equivalent to line
            break on only one of West or East node sides.
The alternative connectivity pattern, called uni-directional, is represented in Figure 7-3 and
is characterized by the fact that:
           every TX-RX tributary uni-directionally transmits to (or receives from) the West
            side and receives from (or transmits to) the East side,
           it could be unbalanced with respect to transmission distances on the two
            propagation directions,
           it might require particular attention to:
               o    alarm propagation,
               o    transmission quality,
               o    time delays management;
           it is neutral for application to protected traffic patterns,
           it is suitable for uni-directional traffic (uni-directional rings and buses),




                  West line ports                                 East line ports




                                          Local ports

           Figure 7-3 – The alternative connectivity pattern for the SC-OADM subsystem.

           a “vertical” grouping of base modules (Figure 7-4) does not guarantee any
            protection in case of fault of one of the modules: this fault is equivalent to a
            simultaneous line break on both West and East node sides;
           a “horizontal” grouping of base modules (Figure 7-5) does guarantee protection
            in case of fault of one of the modules: this fault is equivalent to line break on
            only one of West or East node sides.




                                         Page 79 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




                                        DO                      AE
                         INn                    PTn-1                 OUTn




             West                                                               East
             line                                                               line
                         OUTn                   PTn-1                 INn




                        West Card       AO                DE     East Card


Figure 7-4 – Internal architecture of a bi-directional ROADM subsystem realized interconnecting two
                                 vertical cells via pass through ports.




                                        DO                       AE
                        INn                     PTn-1                  OUTn



                                            W to E card
           West                                                                    East
           line                                                                    line
                        OUTn                    PTn-1                  INn




                               AO        E to W card             DE


  Figure 7-5 – Internal structure of a uni-directional ROADM subsystem realized by two horizontal
                                                  cells.

 7.2     ROADM structure
In order to provide a basic comparison among different approaches, in this section the
common ROADM structures are described.




                                         Page 80 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




Several ROADM architectures have been proposed48, based on many different
technologies, and most of them are yet at research level. In the picture below49 different
possible ROADM architectures are represented. Without the aim to be exhaustive, this
representation is reported to describe how many different approaches have been
considered to implement ROADMs, many of them tied to particular technologies.
Among all these different architectures, some have severe constraints (for instances being
not hitless) and just few have proven to be serious candidates for the implementation of
ROADMs.
Architectures 1-5 reported in Figure 7-6 connect only one output signal to any given output
fiber. Architectures 6-9 integrate multiplexers directly into the switch fabric and enable
wavelength-multiplexed output.




                                                                     48
                         Figure 7-6 – Possible ROADM architectures

Another possible consideration is relevant to the number of add and drop ports: while it is
common to have a balanced architecture, but it is also possible to design asymmetric
architectures50 that exploit the functionality of a specific technology, such as tunable filter
(see Figure 7-7a) or tunable lasers (see Figure 7-7b) without having to use both. For
example, in an access network, where all traffic flows to a hub, a wavelength can be
devoted to every non-hub node, using tunable lasers in the hub to reach the various nodes,
without tunability on the drop side (architecture (b) in Figure 7-7).




                                       Page 81 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




Figure 7-7 – Block diagram of: asymmetric ROADM with Drop tunability a); asymmetric ROADM with
                                                                           50
                       Add tunability b);ROADM with Add/Drop tunability c)

Considering a subset of the possible ROADM architectures, the structures that are
currently evaluated as suitable solutions are:
   -   Mux/demux + 2x2 switch
   -   Mux/demux + NxN switch
   -   Wavelength Blocker
   -   Serial tunable OADM
   -   Wavelength Selective Switch
The following sections also discuss some among available/proposed technologies
proposed for implementing ROADM functions and the basic building blocks required.
Considering commercial available products, pre-production prototypes and promising
research activities, several products are based on tunable filter technologies, so references
are made to the relevant sections, while specific products based on different approaches
are discussed in the following.


 7.3     Mux/demux + 2x2 switch
This ROADM structure (Figure 7-8) is based on the possibility to access one or more sub-
bands inside an OADM node, while leaving other bands to be transparently passed
through. The receive line side band splitter separates the available incoming optical sub-
bands. The sub-band(s) containing the optical channels to be terminated inside the node is
(are) completely de-multiplexed and the optical channels are connected to the client units.
On the transmit side, the output of the client is physically connected to the related ports of
the multiplexer unit, the output(s) of which is (are) then connected to the band combiner on
the line transmit side. Direct connection between band splitter/combiner ports related to
express sub-bands, consents the transparent pass-through traffic.



                                       Page 82 of 144
                                                                  First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                            3ab4c127c833.doc




A selected channel can be A/Ded or leaved on the pass-trough thanks to the 2x2 switch.
This is a general two-stages approach, which can be modified for instance adopting only a
larger mux/demux stage (without band splitter/combiner), with 2x2 switches on all the
channels, so increasing the tunable bandwidth.
It is a quite simple and straightforward approach suitable for high number of A/D ports; it
also allows integrating VOA functions on demultiplexed channels, for lambda conditioning
functions. Nevertheless, the structure is not flexible (a single factory-selected lambda can
be A/D only at a specific port) and the banded approach maintains limitations common to
the fixed OADM structure, in terms of stranded lambdas and ASE accumulation (a special
node in a ring shall be devoted to ASE filtering).
Insertion loss for low channel number (TFF adopted) is around 4-7 dB, while at higher
channel count (AWG adopted) there is a trading between loss and spectral narrowing when
cascading ROADMs and insertion loss is around 12 dB.
The effectiveness of this approach strongly depends on the number of optical sub-bands
available and on the number of optical channels per sub-band.
One drawback is that the stage demultiplexer at sub-band level could require a guard band
between each sub-band without any payload optical channel. This means that part of the
available optical amplification band could be not available for payload channels. Another
point is that every sub-band is composed of narrow-spaced optical channels, and this fact
requires the use of high quality muxes/demuxes at sub-band level for accessing the optical
channels.
The structure does not guarantee East/West separation.




                    Figure 7-8 – Schematic diagram for 2x2 switch ROADM




                                       Page 83 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




                        PROS                                        CONS
         Single lambda access (VOA, OPM) in          Each lambda is linked to a specific
         case of full demux                          A/D port
         Simple and straightforward approach,        Single point of failure (no East/West
         suitable for 100% A/D                       separation)
         ‘Opaque’ node in case of full demux         With sub-band approach: stranded
         (no ASE accumulation in a ring). In         lambdas
         case of sub-band approach, a special
         node in a ring shall be devoted to ASE
         filtering
         Low insertion loss (4-7 dB) in case of      Trading between AWG loss and
         low channel number                          spectral narrowing when cascading
                                                     ROADMs



 7.3.1       Overview of existing products
There are several providers offering products falling in this category, due to the quite
simple architecture and different technologies suitable to implement the needed building
block.
The connectivity performances of all the products adopting this architecture are very
similar, since they do not support the ‘any-lambda-to-any-port’ feature and the only
difference within the products is the tuning range (for this category the tuning range is
proportional to the number of 2x2 switches, typically corresponding to the number of
demultiplexed lambdas).
The optical performances depend on the kind of multiplexer/demultiplexer and possible
band splitter/combiner adopted: in case of few optical channels managed the preferred
solution is based on Thin Film Filters (TFF), while with a high number of channels AWG are
the common choice (even if AWG Gaussian-type are also proposed as an option, the
cascadeability requirement of ROADM normally involves the adoption of flat-top AWG, that
in turns means increase of insertion loss). Channel spacing down to 50 GHz is not an
issue, since suitable multiplexer/demultiplexer are widely available.
The technology of 2x2 optical switch mainly influences the switching time. Thermo-optic
switch modules (the most widely adopted solution) are comprised of 2x2 integrated-optic
Mach-Zehnder interferometers (switch elements), constructed on waveguide material
whose refractive index is a function of temperature. These devices are normally produced
via silica on silicon or using polymeric substrate. They are capable of continuous output
power control, and have a switching time of a few milliseconds.
Solid-state 2x2 switches are ideal in situations where high channel count in a compact
package is of prime concern. Solid-state switches are electrically controlled and have no
moving mechanical parts.
In the table below, main characteristics of publicly promoted products are reported. Since
some architectural and operational characteristics are common (all products are hitless, no
branching is possible, no East/West separation), the only points reported are relevant to




                                       Page 84 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




channel equalization (reported in the ‘VOA’ column) and optical power measurement
(reported in the ‘OPM’ column): Both VOA and OPM only apply to the demultiplexed
channels (i.e. channels belonging to pass-though sub-bands cannot be equalized or
measured). For this type of product, assuming a reference channel spacing of 100 GHz,
the tuning bandwidth can be defined as K*100 GHz (K=number of add/drop ports). Isolation
of adjacent and non-adjacent channels is dependant on the band-splitter and demultiplexer
adopted, since some products simply provide the multi port switch building block.
In the following table, main characteristics of products are reported.




                                       Page 85 of 144
                                                            First Guideline for Technologies
                        IST IP NOBEL "Next generation
                        Optical network for Broadband               ee787c6d-c81c-433e-a3ea-
                             European Leadership"                          3ab4c127c833.doc




            Parameter            Symbol      Unit       Santec        Optun       Dupont        Global      Chromux       Hoya
                                                                                               Opticom
                                                        i-OADM51    A-Block52    iROAD888                   ROADplex      RC-
                                                                                      53
                                                                                               Program        er55       OADM56
                                                                       (**)                     mable
                                                                                     (**)      OADM54
Operating Bandwidth                 BO       nm     C or L band     1525-1570    1528-1610       C-band         -        1470-1610
                                                                                               (1529.55 -
                                                                                                1560.61)
                                                                                                L- band
                                                                                               (1570.42 -
                                                                                                1603.17)
Tuning Bandwidth                    BT       nm            3.2         3.2           6.4           6.4        12.8          3.2

A/D ports                           K         -         4 (8, 16)       4             8        (4) 8 (16)       16          4
                                                                                                            (20,32,40)
Channel Spacing                            GHz        100 (50)        (*)          (*)          100          100           -

Tuning Time                         TT       ms            10           10           10           10            3           10


Pass Through Loss of non-          PT       dB           <3.5         2.5           2.1           7           11           3
dropped channels (express                                express
channels)                                                  < 6.5
                                                         through
Pass Through Loss Uniformity       UPT       dB             -            -            -           1.2          0.1           -
of non-dropped channels
(express channels)




                                                           Page 86 of 144
                                                            First Guideline for Technologies
                       IST IP NOBEL "Next generation
                       Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                            European Leadership"                           3ab4c127c833.doc




           Parameter             Symbol     Unit       Santec         Optun       Dupont        Global      Chromux       Hoya
                                                                                               Opticom
                                                       i-OADM51     A-Block52    iROAD888                   ROADplex      RC-
                                                                                      53
                                                                                               Program        er55       OADM56
                                                                       (**)                     mable
                                                                                     (**)      OADM54
Rejection of dropped               CIB      dB             -            (*)          (*)           -             -         -
channels
Drop Loss                          d       dB           < 4.4           -           1.7          5.5           6         2.5

Drop isolation of the adjacent    COBda     dB             -            (*)          (*)          30            25         -
channels
Drop isolation of non adjacent   COBdna     dB             -            (*)          (*)          45            32         -
channels
Add Loss                           a       dB            <4             -           2.0          5.5           6         2.5

Add isolation of the adjacent     COBaa     dB             -            (*)          (*)          25            25         -
channels
Add isolation of non-adjacent    COBana     dB             -            (*)          (*)          45            32         -
channels
Filter Central Wavelength         prec     nm             -             -            -         0.1 nm       +/- 0.015     -
Accuracy                                                                                                        nm

Filter Pass-band                   BC       GHz         +/- 12.5         -            -        ± 15 @ 0.5   50 @ 1 dB      -
                                                                                                   dB
                                                                                               ± 30 @ 3.0
                                                                                                   dB




                                                           Page 87 of 144
                                                                 First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                           3ab4c127c833.doc




               Parameter               Symbol       Unit     Santec        Optun        Dupont        Global       Chromux    Hoya
                                                                                                     Opticom
                                                             i-OADM51     A-Block52    iROAD888                   ROADplex    RC-
                                                                                            53
                                                                                                     Program        er55     OADM56
                                                                             (**)                     mable
                                                                                           (**)      OADM54
     Polarisation Dependent Loss         PDL        dB         < 0.2        < 0.5          0.5          0.15            1     0.3

     Pass-band ripple                              dB           -            -             -            0.5            -      -


     VOA                                 VOA         -         Yes           Yes           Yes            -            Yes     -

     OPM                                 OPM         -         Yes           Yes           Yes            -            Yes     -


(*) = mainly dependant on external band-splitter and/or demultiplexer
(**) = switch module only, so multiplexer/demultiplexer characteristics shall be added (insertion loss, bandwidth,…)




                                                                Page 88 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




The table represents just a sample of the proposed products that are reported below: it is
apparent that the main characteristics of products belonging to this category are aligned.
Santec iOADM is based on Thin-Film-Filter demultiplexer and opto-mechanical switch51.
Optun A-Block Optun patented Controlled Mode interaction (CMI) is a Silicon-based PLC
technology adopted to implement wideband, low cross-talk, flat-response 2x2 switches,
based on SiO2 on Si waveguides and thermal switching. The iOADM product is a 4-port
configurable add/drop module without mux/demux52.
Dupont iROAD888 The iROAD888 product is an 8-port configurable add/drop module
without mux/demux, based on integration on polymer-based PLC53.
Global Opticom (Aoctech) Programmable OADM A similar product is also promoted by
Aoctech. It is based on TFF band-plitting/demultiplexer and opto-mechanical switch54.
Chromux ROADplexer MEMS-based switch55. The product should integrate the C-MEMS
technology developed by NP Photonics, described in the section XXX relevant to tunable
filters.
Hoya RC-OADM Is based on a TFF demultiplexer with micro-optic switch (two moving
collimater and micro solenoid-coil with latching configuration). Optical equalisation with
SOA option56.
Neophotonics ROADM PLC-based product, with AWG demultiplexer
JDS COADM (Configurable Add/Drop Multiplexing) The standard JDS Uniphase
COADM uses 2x2 opto-mechanical Switch Modules. These switches use reconfigurable
optics, combining gradient index (GRIN) lenses and polished mirrors operated by a reliable
electro-mechanical relay assembly. The switches have latching capability.

 7.3.2      Existing technologies and further developments on basic building
            blocks
PLC switches can be implemented with basing on solid state elements (exploiting thermo-
optic or electro-optic effect), with MEMS or fluids.



   7.3.2.1 PLC and MEMS combination
A method to implement optical switch arrays and VOA arrays with 2x2 cross type
waveguides and MEMS micro-mirrors (connected via flip chip bonding, as depicted in
Figure 7-9) have been proposed57.




                                       Page 89 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




                                                                        57
                        Figure 7-9 – Structure of PLC and MEMS device

The operating principle is described in Figure 7-10. In (a) the input signal is reflected by the
micro-mirror (bar state). In (b) the micro-mirror is moved along the trench up to the cross
state by applying a voltage (the cantilever beam is moved in parallel with the PLC surface
and the micro-mirror is moved along the trench). Thus the device, with an On/Off voltage
applied, behaves as an optical switch (switching time is about 4 msec), while when the
voltage is varied in an analog way, the device operates as a VOA.
The insertion loss is < 1.0 dB, the PDL < 0.1 dB both in the cross state and in the bar state.
The cross-talk < -50 dB. The wavelength dependent loss over the entire 1520 - 1610 nm(C
and L band) range is < 0.1 dB.
The package for the 12 channel array is 50mm(L) x 20mm(W) x 6mm(T).




                                                                                  57
               Figure 7-10 – Schematics of the multichannel optical components.




   7.3.2.2 PLC integration of ROADM
Efforts in the direction of high integration of ROADMs on PLC platforms have been made
and several efforts have been made on the mux/demux + 2x2 switch structure thanks to
the simplicity of its basic building block.
Several design were proposed based on AWG technology for the multiplexer/demultiplexer
and this benefit from technological progresses in improving optical features (insertion loss,
polarization dependence) of this elements, widely used in several DWDM applications.
Studies on AWG are also focusing on the flat response required in order to allow ROADM



                                        Page 90 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




cascadeability. Methods to achieve flat-top bandpass, with respect to standard Guassian-
type bandpass can be based on spatial filtering, on waveguide grating combined with
interleaver or on synchronized waveguide grating58.
Design of PLC integrated ROADM based on this architecture started in mid nineties and
several different design and technology were proposed, also following AWG improvements.
Later, efforts were more focused on integration of other functions such as VOA for
equalization and 2x2 switch. As a recent example, a ROADM on a chip with 20 channels at
200 GHz spacing can be reviewed59.
The 20 2x2 switches as well as the VOAs employ multiple stages of standard, Mach-
Zehnder (MZ)-type thermo-optical elements. Thecnological methods60 allow to reduce both
the power consumption of each single thermo-optic element (well below 100 mW) and the
thermal crosstalk.
The photodiodes for power monitoring are preassembled and tested on a glass submount
in units of 10 diodes and are then flip-chip mounted onto the PLC motherboard.
The ROADM optical chip occupies a 4” wafer and showed a pass-through insertion loss of
9 dB, with a switch extinction ratio >40 dB.




                                                                                            59
        Figure 7-11 – Silica on Silicon ROADM structure (left) and implementation (right)

These studies were carried out by Infineon; later, Infineon sold part of its business to
Optun.



   7.3.2.3 Polymer based ROADM
Polymer-based PLC have been introduced and Dupont Photonics has developed a
platform including switches.




                                        Page 91 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




An interesting way to overcome the lack of East/West separation that is one of the
limitations of this ROADM architecture, an alternative approach can be to substitute a 2x2
switch with a couple of 1x2 switches. A similar approach, on integrated polymer-on-silicon
platform, has been proposed by Dupont61. This solution (Figure 7-12) is based on two
individually packaged chips, including an array of 1x2 switches, as well as VOA and power
tap; strictly non-blocking 8x8 switches allow full reconfigurability (i.e. ‘any-wavelength-to-
any-port’) and can be considered an hybrid between mux/demux + 2x2 and NxN switch
architectures.
Elements demonstrated on this polymer-on-Silicon platform range from 2x2 switches with
high isolation (57 dB) and low power consumption (66 mW), waveguide with 0.11 dB/cm
propagation loss, VOA (MZI-based) with 30 dB isolation at 1.4 mW of power consumption
and 8x8 switch composed by cascaded 1x2 switch with 2.9 dB insertion loss.




   Figure 7-12 – Wavelength agile 8-channel ROADM without East/West separation (a) and with
                                                            61
                                   East/West separation (b)

Recent developments62 encompass the hybrid integration of AWG PLC as
multiplexer/demultiplexer block with polymer-based PLC devoted to switching (array of
independent 1x2 switches), achieving 40 channels full reconfigurability with East/West
separation and with a minimum of 4 dB of insertion loss on express channels (Figure 7-13).




                                       Page 92 of 144
                                                                     First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                              3ab4c127c833.doc




  Figure 7-13 – Layout of a 40-ch ROADM subsystem, comprising two modules, each including a
      polymer PLC, chip-to-chip attached to silica AWGs. The subsystem features Add/Drop
  functionalities, automatic load balancing and power monitoring. The 64x64 OXC functionality is
                                                                    62
                                realized in a separate polymer PLC.

Among other activities on polymer devices based on a similar structure, HHI (Heinrich-
Hertz-Institut) and Fraunhofer Institute for Reliability and Microintegration presented63 an
implementation of a four channels OADM (two AWGs and a 2x2 DOS array) with
Fluoroacrylate polymer, obtaining 30 dB crosstalk.

   7.3.2.4 Pass-band filters
Some ROADM architectures adopt bands of channels to facilitate add/drop multiplexing, so
involving the usage of bandsplitting or “skip” filters.
One common technology to implement passband filter is based on Thin Film Filter (TFF).
An example of a bandsplitting filter is the 8-skip-2 filter for 100 GHz-spaced channels,
which transmits 8 channels and achieves isolation between the transmitted and reflected
channels by skipping two channels. Since skipping channels is undesirable, as it has the
effect of reducing the available communication bandwidth, filters with steeper slopes such
as the 8-skip-1 filters, which skips only one channel, and 8-skip-0 and 4-skip-0 filters have
been developed64.
Due to the required steeper transitions between passbands and stopbands, the filter design
adopt more cavities, so resulting in thick structures and consequently to film stress (that in
turn generates a variation in filter center passband wavelength). Thus research efforts were
focused on development of special low-stress process.
Other major points in TFF bandsplitting filters is the reduction of ripple and CD: passband
filter 4-skip-0 (the multicavity desing is reported in Figure 7-14) with CD of 19 ps/nm and
ripple of 0.07 dB have been demonstrated65.




                                        Page 93 of 144
                                                                        First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                        ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                                   3ab4c127c833.doc




                                                                         65
                        Figure 7-14 – Structure of TFF passband filter

Since many of ROADM devices are developed in PLC technology, methods to integrated
passband filters have been investigated (in a 8-skip-1 filter shape is reported)66.




                                                  66
          Figure 7-15 – 8-skip-1 band mux/demux        (in collaboration with Neophotonics)

Among proposed products enabling 8-skip-0 and 10-skip-0 band-splitting on a 100 GHz
grid, Bookham introduced its TFF-based (exploiting Bookham's proprietary AED III -
Advanced Energetic Deposition, version III - technology,) flexible upgrade architecture for
high channel count, scalable mulitplexer/demultiplexer solutions, suitable to be upgraded in
8-10 channel increments in any wavelength order.
Figure 7-16 shows 8-skip-0 architecture and the spectral response of the filter67.




                                       Page 94 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                        ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                   3ab4c127c833.doc




                                                                              67
                   Figure 7-16 – Example performances 8-skip-0 architecture


 7.4     Mux/demux + NxN (NxM) switch
This architecture (Figure 7-17) is similar to the previous one but the improvement in
flexibility given by the NxN switch (with N double of the number of A/D ports) allows
choosing any A/D port for a selected channel.
The architecture features a band-splitter that divides the channels grid into separated sub-
bands. One of them is demultiplexed in separated channels, which can be individually
dropped by means of an NxN switch fabric. The express and the added channels are then
multiplexed again with all the sub-bands through a band-combiner as shown in the picture.
Different versions (with MxN switch and without sub-band approach) can also be
implemented.
Basically similar considerations to the 2x2 switch structure apply, with the only difference
that a certain degree of flexibility is provided (but at the cost of a large switch).




                                       Page 95 of 144
                                                                 First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                           3ab4c127c833.doc




                    Figure 7-17 – Schematic diagram for NxN switch ROADM

                        PROS                                      CONS
         Single lambda access (VOA, OPM) in        Not-blocking A/D feature requires
         case of full demux                        large switches
         Simple and straightforward approach,      Single point of failure (no East/West
         suitable for 100% A/D                     separation)
         ‘Opaque’ node in case of full demux       With sub-band approach: stranded
         (no ASE accumulation in a ring). In       lambdas
         case of sub-band approach, a special
         node in a ring shall be devoted to ASE
         filtering
         Low insertion loss (4-7 dB) in case of    Trading between AWG loss and
         low channel number                        spectral narrowing when cascading
                                                   ROADMs


 7.4.1       Overview of existing products
Lynx Photon.Net 8/8 Lynx Photonic Networks developed an 8 x 8 array based on Mach-
Zehnder switch structures. The MZI switching operation is based on the thermo-optic effect
that provides analog phase-shifting, enabling variable-output power ratios. The switch’s
ability to manipulate phase enables it to perform extra switching functions such as
weighted multicasting (switching of single signals to more than one output) and dynamic
power management, including attenuation and integrated spectrum analysis.



                                       Page 96 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                      ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                 3ab4c127c833.doc




Photon.8x8 PCSSTM , a strictly non-blocking photonic switch with 8 input ports and 8 output
ports, is a subsystem of a product family with different modularity. It shows low loss (<3
dB), low crosstalk (<50 dB) and 2 msec switching time68.
Dupont iROADTM 888 is a product that, basing on a combination of 1x2 and 8x8 switches,
allows East/West separation and any-lambda-to-any-port accessibility, as described in
previous chapter. Dupont iOXCTM 800 is a strictly not-blocking and hitless 8x8 optical
switch69.

 7.4.2      Existing technologies and further developments on basic building
            blocks
Similar to the ROADM structure based on 2x2 switch, studies on integration of structures
based on NxM switches are ongoing.

   7.4.2.1 PLC integration
A monolithically integrated Silica-on-silicon (silica-based) planer lightwave circuit (PLCs) 32
x 4 ROADM has been demonstrated70.
The 32x4 ROADM (Figure 7-18) integrated all the required components monolithically,
including two array waveguide gratings (AWG) for multiplex/demultiplex and a 37x4 (37 is
the number of wavelengths and 4 is the number of add/drop ports) crossbar switching
fabric.




                                                                            70
                    Figure 7-18 – Structure of client-reconfigurable OADM

As can be seen from the picture, all demultiplexed channels are sent to a crossbar switch
matrix for add/drop operation: each drop wavelength can be directed to any of the 4 drop
ports and the corresponding add signal, of the same wavelength, is then added to the
system. All the output channels are then multiplexed by a second AWG.
The crossbar switch is implemented using 2×2 switches (with default state in “through”
status). To drop the wavelength j (j = 1 to 32) to port i (i=1~4), the corresponding cross-



                                       Page 97 of 144
                                                                         First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                       ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                                  3ab4c127c833.doc




point j-i is switched on, that is, the 2×2 switch is set to “bar” status. At the same time, a new
signal of wavelength j can be added into port i.
The 2x2 switch can introduce first order crosstalk on the signal to be added/dropped:
reducing it to second order crosstalk is accomplished by dilating the switching unit to avoid
two active input ports for any 2×2 switching unit. The dilated design of the cross-point
structure is shown in Figure 7-19 (full dilated (a), simplified (b))




    Figure 7-19 –Dilated 2×2 switching unit: (a) fulldilated 2×2 switching unit; (b) simplified 2×2
                                                                      70
                             switching unit for cross-point structure

Figure 7-20 represents the spectrum response of the ROADM under pass status (power
off), showing a power off throughput uniformity better than 2.5dB. (mostly due to power
leakage of the switching units which can be reduced by thermally-trimming the imbalance
of the thermo-optic switch). Figure 7-21 shows the case when one of the channels is
dropped.
The maximum crosstalk level of all pass channels is less than -35dB (the performance can
be attributed to the dilated switching array).
The average fiber-to-fiber loss of pass through channel can be estimated in11~13.5dB,
while the loss figure for add/drop channel will be 8~11.3dB. The measured values reported
were higher (24 dB pass-through and 42 dB drop), due to errors in junctions of bending
waveguides.




                                                                             70
                       Figure 7-20 –Transmission curve for all-pass status




                                          Page 98 of 144
                                                                 First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                               3ab4c127c833.doc




                                                                              70
                Figure 7-21 –Transmission curve when one channel is dropped

The ROADM device total size is 119mm×114mm and the final layout mask can be seen in
Figure 7-22.




                                                                          70
                 Figure 7-22 –Mask layout of a PLC-based integrated ROADM

Further development on this architecture has been proposed71, with a modular structure
(Figure 7-23) allowing capacity expansion and supporting different configurations as ring
interconnection.




                                      Page 99 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




   Figure 7-23 – Modular OADM design (left) and cascaded 2 ROADMs for doubling the channel
                                                        71
                                       capacity (right)




   7.4.2.2 PLC NxN Switches
Silica-based PLC switches have various advantages, such as low insertion loss,
polarization insensitive operation, fabrication repeatability and long-term stability. One of
the issues for this kind of component, especially for large scale switches, is the power
reduction, that is normally achieved using heat-insulating grooves. Typical switching
response time is of few milliseconds.
The basic switch element is a balanced bridge Mach-Zehnder interferometer (MZI) with thin
film heaters that act as phase shifters and it operates by employing the thermo-optic effect.
A strictly non-blocking NxN matrix switches with a scale of up to 16x16 has been
demonstrated72; the basic building block is a double-MZI switching unit (to provide a high
extinction ratio) is represented in Figure 7-24 (a), while the logical arrangement to
implement an NxN switch module is represented in Figure 7-24 (b).
A 16x16 switch module with the switch chip and a driving circuit integrated has insertion
loss of 5.6 dB, extinction ratio of 60 dB and power consumption of 13 W.
As an additional function that can be integrated in the same platform, VOA array allow to
adjust optical power level of the managed channels. An array of 16 VOAs has been
demonstrated using asymmetric MZI obtaining an average insertion loss of 1.03 dB and a
PDL below 0.5 dB73.




                                       Page 100 of 144
                                                                       First Guideline for Technologies
                                IST IP NOBEL "Next generation
                                Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                     European Leadership"                               3ab4c127c833.doc




                                                                                                   72
   Figure 7-24 –NxN matrix switch configuration: (a) 2x2 switching unit, (b) logical arrangement


  7.4.2.3 Polymer-based PLC NxN switch
Dupont has developed a polymer-based PLC technology including switches. 8x8 switch74 is
implemented with a scheme reported in Figure 7-25, which gives some advantages with
respect to traditional approaches.
It operates in C or L band in a hitless way and it is able to perform multicast and broadcast
routing. It is based on 112 1x2 switches, with power consumption below 3 W.
Insertion loss is below 4.5 dB and options include the path independent optical loss,
extinction of 45 dB and switching time of 3 msec.




                 Figure 7-25 –8x8 matrix switch obtained cascading 1x2 switches



                                         Page 101 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                             3ab4c127c833.doc




 7.5     Wavelength Blocker
This ROADM architecture (Figure 7-26) is based on a component (Wavelength Blocker)
able to selectively block (or attenuate) selected lambdas (one, several or all) on the
express path.
A/D is provided by means of splitter/coupler with mux/demux (or coupler/decoupler with
tunable filters). Insertion loss for Wavelength Blockers is typically around 6 dB at which
coupling losses should be added on the express path
This structure has good performances in terms of network-level features (East/West
separation, off-ring upgrades not affecting express traffic, scalability and pay-as-you-grow,
possibility to equalize the optical power of all the channels, inherent prevention of ASE
accumulation), but has currently an high upfront cost compared with previous solutions
and, in the basic version with mux/demux, does not provide flexibility since each lambda
can be A/Ded only at a specific port.
To achieve full flexibility, a splitter (combiner) should be adopted instead of the demux
(mux), so requiring tunable filters at receivers (and with high insertion loss on through-drop
and add-through paths as A/D ports increase).
A Wavelength Blocker–based structure could implement branching unit functionality.




                    Figure 7-26 – Schematic diagram for Wavelength Blocker

                        PROS                                        CONS
       Pay-as-you-growth,      with    off-ring     High upfront cost
       upgrades
       East/West separation                         ‘Not    blocking    OADM’      requires



                                      Page 102 of 144
                                                                  First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                           3ab4c127c833.doc




                                                   tunable filters and splitters (high IL
                                                   with high channel count)
         Suitable for 100% A/D
         Single lambda management (VOA,
         OPM)
         ‘Opaque’      node      (no      ASE
         accumulation)



 7.5.1       Overview of existing products
In the following, an overview of publicly promoted products is provided: it should be noted
that Wavelength Blockers have several possible applications, one being ROADM and to
build a Wavelength Blocker based ROADM other components must be integrated.
Thus, in the following table, some of the products are compared basing on typical
Wavelength Blocker parameters and not basing on ROADM subsystem parameters.




                                       Page 103 of 144
                                                                  First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                           3ab4c127c833.doc




         Parameter                 Unit       Santec           JDSU         Avanex      LightCon      Polychro     Xtellus
                                                                                          nect          mix
                                             LBM-875       WB 50 GHz        Power                                   DBE80
                                                            C-band76       BlockerTM      DCE    78
                                                                                                        P-
                                                                               77
                                                                                                      DCOTM79
Wavelength Range                    nm        C/L band         1527.5 to    C-band       C-band        C-band      C-band
                                                              1565.6 nm
                                            4-16 ch max.                                              (100 chs)    (126 chs)
Insertion Loss                      dB           7                7            6            6             6           5

Channel Spacing                    GHz        100, 200           50           100           50           50           50

Response Time                       ms          10               30            -           0.1            1            -

Attenuation Range                   dB          0-20            0-20           20          0-10         0-15         0-15

Channel Blocking                    dB          45               32            40           40           40           40

Passband                           GHz        25 GHz       +/-10GHz @       25 GHz       25 GHz       +/- 12.5 @       -
                                              @0.5dB          0.5 dB        @0.5dB                      0.5 dB
Polarisation Dependent Loss         dB          0.4              0.5          0.5          0.3           0.3           -

Passband Ripple                     dB           -               0.3          0.3            -            -            -

Chromatic Dispersion              ps/nm          -                -          +/- 10        +/- 3          -            -




                                                                 Page 105 of 144
                                                                   First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                 ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                            3ab4c127c833.doc




Santec LBM-8 LambdaBlocker
It is a product based on demultiplexing the channels (100 or 200 GHz, via TFF filters), then
equalizing/blocking each single channel with proprietary MEMS technology (0 to 20 dB
attenuation) and the multiplexing channels again. Photo-Detectors for optical power
monitoring can be optionally included75.
JDS Wavelength Blocker
It is a 50 GHz and 100 GHz grid product. Different versions can manage C-band and L-
band76.
Avanex Power BlockerTM
It is a 100 GHz product (two versions: one for C-band and the other for L-band) based on
proprietary technology with a high extinction ratio77.
CoAdna DCE
The DCE (Dynamic Channel Equaliser) and Wavelength Blocker module is base on liquid
crystal technology, with 50 and 100 GHz channel spacing.
LightConnect DCE
It is a 50 GHz product, able to manage channels across C-bands, based on diffractive
MEMS technology78. Details about this technology are reported in section 7.5.2.
Polychromix Dynamic Channel Orchestrator P-DCOTM
It manages 100 channels at 50GHz or 100 GHz channel spacing (the table reports data
relevant to 50 GHz channel spacing) and uses MEMS programmable micro-diffraction
gratings, its special characteristic being that the MEMS surfaces managing light are flat79.
Details about the technology are reported in section 7.5.2.
Silicon Light Machines SLM 3000 Reconfigurable Blocking Filter
SLM3000 product is based on patented Grating Light Valve™ (GLV™) MEMS technology,
acting as a tunable grating to vary the amount of laser light that is diffracted or reflected.
Details on the technology are reported in section 7.5.2.
Xtellus Dynamic Blocker Equalizer
DBE (Dynamic Blocker Equalizer) is a liquid crystal based device able to support 126
channels at 50 GHz spacing or 63 channels at 100 GHz spacing80.

 7.5.2      Existing technologies and further developments on basic building
            blocks
LightConnect adopts Diffractive MEMS (D-MEMS) technology as the platform for
implementing voltage-controlled variable-diffraction grating.
The D-MEMS actuator consists of an array of reflective ribbons suspended above a
reflective substrate. As depicted in Figure 7-27, when no voltage is applied the height
between the ribbon array and the substrate is one wavelength, so the light reflected from
these two surfaces is in phase and the device functions as a mirror. A voltage applied
between the ribbon array and the substrate creates an electrostatic force that flexes the
ribbon array, therefore decreasing the height and creating a phase difference between the
light reflected from the two surfaces81.




                                      Page 106 of 144
                                                                      First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                    ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                               3ab4c127c833.doc




                                                                                                    81
 Figure 7-27 – The D-MEMS actuator functions as a voltage-controlled variable diffraction grating

In the D-MEMS CMOS process, a layer of polysilicon is deposited on the silicon substrate,
then ribbons are fabricated of silicon nitride using normal lithographic processes. Once
ribbons are defined, the polysilicon is removed and a layer of aluminum is applied across
all surfaces, which coats both the ribbons and the surface of the substrate between ribbons
with a reflective material (Figure 7-28 –D-MEMS is manufactured using 1m CMOS
process81).




               Figure 7-28 –D-MEMS is manufactured using 1m CMOS process
                                                                                  81




Polychromix technology is based on electrically programmable diffraction grating, which
consists of an array of parallel reflective gold-coated polysilicon mechanical beams whose
vertical position can be independently controlled (represented in Figure 7-29). This device
can be combined with fixed optical elements to create a dynamic spectral equalizer, either
on a channel-by-channel or banded basis, and to create a programmable channel- blocker
for use in OADM and wavelength-based switching applications82.




                                        Page 107 of 144
                                                                          First Guideline for Technologies
                                  IST IP NOBEL "Next generation
                                  Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                       European Leadership"                                3ab4c127c833.doc




One remarkable characteristic is that Polychromix diffractive device has resonant
frequencies well outside the range of vibration or shock sensitivity, and has only one
degree of freedom of actuated motion, tightly constrained by element supports.




 Figure 7-29 –A packaged Polychromix die, consisting of an array of suspended parallel polysilicon
                                                                   82
                           beams (left), as shown enlarged (right)



Silicon Light Machines bases its DCE product on patented technology. Grating Light
ValveTM (GLVTM) is a diffractive MOEMS spatial light modulator capable of high-speed
modulation of light combined with fine gray-scale attenuation.
The device (Figure 7-30) is built on a silicon wafer and is comprised of many parallel micro-
ribbons that are suspended over an air gap above the substrate83.




                          TM
   Figure 7-30 –The GLV        Device with alternate ribbons deflected to form a dynamic diffraction
                                                        83
                                                grating

The diffraction grating is implemented deflecting the electrically conductive ribbons toward
the substrate; if no voltage is applied, ribbons are coplanar so, being highly reflective, they
act as a mirror and incident light is reflected. When alternate ribbons are deflected, the




                                           Page 108 of 144
                                                                       First Guideline for Technologies
                               IST IP NOBEL "Next generation
                               Optical network for Broadband                     ee787c6d-c81c-433e-a3ea-
                                    European Leadership"                                3ab4c127c833.doc




angular direction in which incident light is steered from the device is dictated by the spatial
frequency of the diffraction grating formed by the MOEMS ribbons (Figure 7-31).




Figure 7-31 –Cross-section of the GLV device, wr is the ribbon width, wg the inter-ribbon gap width,
                                                                                              83
       h the ribbon-to-substrate height, Λ the diffraction period, and δ the ribbon deflection .

Ribbons are 1-10µm wide, 200-300 nm in thickness and 100-1000µm long, with an
aluminum top layer (acting both as reflective layer and as top electrode for electrostatic
actuation). The sub-layers of the ribbon are a sandwich of stoichiometric Si3N4 and SiO2
films, devoted to provide the restoration force that counter-balances the electrostatic
actuation force, and stiffness and stress balance so the ribbon remains flat across its width.
A complete device is implemented replicating ribbons (closely spaced, with gap around
0.5µm) several thousand times to form a 1-dimensional array of diffracting elements
(Figure 7-32).




                                                                        83
                         Figure 7-32 – SEM photograph of GLV ribbons




                                        Page 109 of 144
                                                                    First Guideline for Technologies
                              IST IP NOBEL "Next generation
                              Optical network for Broadband                      ee787c6d-c81c-433e-a3ea-
                                   European Leadership"                                 3ab4c127c833.doc




Liquid Crystals
Xtellus adopts liquid crystals to develop a platform (Liquid Crystal Optical Processor) for
optical communication devices that switch, filter and attenuate light, among them the
Dynamic Block Equalizer.
A simple schematic of a liquid crystal processor is illustrated in Figure 7-33. Liquid crystal
material is embedded between two pieces of glass that is coated with transparent
conductive indium tin oxide (ITO) to create the active pixel elements. Electrical leads are
terminated on the ITO defined pixels84.
By applying electric voltage, power level and polarization state can be modified for each
pixel, so influencing the passing-through light beams (with an insertion loss around 0.1 dB).




                                                                            84
                     Figure 7-33 –Liquid Crystal Optical Processor scheme


 7.6     Serial Tunable OADM
This product category (Figure 7-34) includes a wide range of devices, with many different
technologies. Insertion loss is generally low and products currently proposed have a limited
number of ports.
Characteristics and pros/cons heavily depend on technology and specific implementation.
Serial TOADM are anyway specifically designed to fit in new generation networks, so
generally the maximum flexibility is provided and switching time is normally around few
milliseconds.
It should be noted that Tunable OADMs could be generally implemented also in a ‘vertical’
arrangement, so providing East/West separation.
Depending on the technology, the capability to A/D a lambda set could also be
implemented, for branching unit operation.




                                       Page 110 of 144
                                                                  First Guideline for Technologies
                             IST IP NOBEL "Next generation
                             Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                                  European Leadership"                              3ab4c127c833.doc




                   Figure 7-34 – Schematic diagram for Serial Tunable OADM

Due to technology differences on serial tunable products, it is not possible to perform a
comparison with other solutions that is valid for all the components of this category. Main
similarity are the low number of add/drop ports (many product just have a single add/drop
port) and a reduced insertion loss on the through path.

 7.6.1     Overview of existing products
In the following, an overview of publicly promoted products is provided.




                                      Page 111 of 144
                                                               First Guideline for Technologies
                        IST IP NOBEL "Next generation
                        Optical network for Broadband                   ee787c6d-c81c-433e-a3ea-
                             European Leadership"                              3ab4c127c833.doc




            Parameter                 Symbol            Unit           Lambda             Optoplex         Ondax                AFOP
                                                                       Crossing
                                                                                         NovaTunable      Single              SODAlxxxviii
                                                                                     T
                                                                     LambdaFlow           OADMlxxxvi   Channel Filter
                                                                          Mlxxxv
                                                                                                        SwitchTMlxxxvii
Operating Bandwidth                     BO              nm             C or L band           36          1528.77 to       1525 to 1620
                                                                                                          1565.50
Tuning Bandwidth                        BT              nm                 6.4               36          1528.77 to       1525 to 1620
                                                                                                          1565.50
A/D ports                                K               -                  2                 1               1           1

Channel Spacing                                       GHz                50              100, 200          100          100

Tuning Time                             TT              ms                 20               2000             15           20

Routing Power                           RP               -                  1                 1               1           1


Pass Through Loss of non-               PT             dB                  4                1.6              1           1.5
dropped channels (express
channels)
Pass Through Loss Uniformity of         UPT             dB                  -                0.3              -           -
non-dropped channels (express
channels)
Rejection of dropped channels           CIB             dB                  -                25               -           25

Drop Loss                               d              dB                  6                2.5              1           1.8




                                                             Page 112 of 144
                                                              First Guideline for Technologies
                       IST IP NOBEL "Next generation
                       Optical network for Broadband                  ee787c6d-c81c-433e-a3ea-
                            European Leadership"                             3ab4c127c833.doc




           Parameter                 Symbol            Unit           Lambda           Optoplex          Ondax                AFOP
                                                                      Crossing
                                                                                     NovaTunable        Single              SODAlxxxviii
                                                                    LambdaFlowT       OADMlxxxvi     Channel Filter
                                                                         Mlxxxv
                                                                                                      SwitchTMlxxxvii
Drop isolation of the adjacent        COBda            dB                 20               25              25           25
channels
Drop isolation of non adjacent        COBdna           dB                 30               35              35           40
channels
Add Loss                                a             dB                  4                -               -           1.5

Filter Central Wavelength              prec           GHz              1 GHz               -           +/-3GHz         -
Accuracy
Filter Pass-band                       BC              GHz             17@3dB         +/-15 @0.5dB     30 @ 0.5dB       -
                                                                      50@25dB
Polarisation Dependent Loss           PDL              dB                 0.5              0.2             0.2          0.1

Pass-band ripple                                      dB                  -               0.2              -           -




                                                         Page 113 of 144
                           IST IP NOBEL "Next generation
                                                              D20: First Guidelines for Technologies
                           Optical network for Broadband
                                                               Enabling broadband optical networks
                                European Leadership"
                                                                          ee787c6d-c81c-433e-a3ea-
                                                                                 3ab4c127c833.doc




Lambda Crossing LambdaFlow
Is a 2 add/drop channels device, integrating several PLC-based tunable filters (thermo-
optic tuning)lxxxv.
Optoplex NovaTunableOADM
It is based on a variable (dynamically selectable) channel taken from the ITU 200GHz,
100GHz, or 50GHz grid. This device uses proprietary optical processing technology
implemented with micro-optics (micro-actuator based tuning mechanism)lxxxvi. Band
add/drop OADM (100 GHz channel spacing, 2 skip 0 and 4 skip 1) are also available.
Ondax Single Channel Filter SwitchTM
It is available for 100 and 50 GHz channel spacing and is based on 3-D Bragg Gratinglxxxvii,
enabling high isolation add-drop filters (the same filter can be accessed twice by using a
pair of dual-fiber collimators at different depths)
Information on 3-D Bragg Grating can be found in section XXX relevant to tunable filter
technologies. Data reported in the table are relevant to 100 GHz channel spacing version.
Alliance Fiber Optic SODA
It is a single channel, switchable Optical add/drop module (SODA) using patented v-
collimator thin film filter technologylxxxviii.



 7.6.2     Existing technologies and further developments on basic building
           blocks
There are several different designs for serial tunable OADMs, mainly based on innovative
tunable filter technologies: reference is made to the relevant section XXX for an overview
of main development in this area.


 7.7     Wavelength Selective Switch
Wavelength Selective Switch (WSS) is a potential building block for several ROADM
architectures (a possible example in Figure 7-35: it should be noted that the Add section
could also be implemented with a proper WSS). One inherent characteristic of this
structure is that an upgrade to implement higher order nodes (more than just East and
West sides) is generally easier than with the other structures discussed above.




                                      Page 114 of 144
                           IST IP NOBEL "Next generation
                                                                D20: First Guidelines for Technologies
                           Optical network for Broadband
                                                                 Enabling broadband optical networks
                                European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




               Figure 7-35 – Schematic diagram for Wavelength Selective Switch

The conventional architecture for WSC is basedlxxxix on 2x2 switches: a Wavelength
Selective 2x2 Cross-connect is depicted in (Figure 7-36).




                                                                               lxxxix
                Figure 7-36 – Structure of a Wavelength Selective 2x2 Switch

Methods to integrate WSS (with conventional structure or with broadcast and select
structure and with different technologies) have been demonstrated during the past yearsxc.




                                      Page 115 of 144
                            IST IP NOBEL "Next generation
                                                                  D20: First Guidelines for Technologies
                            Optical network for Broadband
                                                                   Enabling broadband optical networks
                                 European Leadership"
                                                                               ee787c6d-c81c-433e-a3ea-
                                                                                      3ab4c127c833.doc




  7.7.1.1 WSS 1xN
An example of integrated WSS 1xN, with eight wavelengths, spaced by 200 GHz has been
demonstratedxci.
The WSS (depicted in Figure 7-37 with a four channels example) is able to extract any of
the eight wavelengths can be removed from the input comb and can be sent to any of the
eight “drop” ports, thus any combination of channels can appear at any drop port. The
undropped channels exit the device at the “thru” port. The device can control the
attenuation of each channel individually at any port by mean of a VOA placed just before
the multiplexer. The add channels optical power level is controlled by means of variable
optical attenuators (VOAs) before combining them with the pass-through traffic.




                                                                                              xci
      Figure 7-37 – Block diagram of the channel-dropping filter for a four-channel example

This structure has some interesting features, besides the ability to send multiple channels
to the same drop port. Every path has at least double leakage rejection in both space and
wavelength, thus requiring just 20 dB of crosstalk suppression on each switching
component to achieve a crosstalk suppression of 40 dB.
Moreover, the proposed architecture is inherently smaller than classic split-and-combine
arrangement of spatial cross connect (less components) and with lower power
consumption (less simultaneously activated switches).
The device layout is shown in Figure 7-38 (with at the bottom the 1x2 switch): as can be
seen the device includes 64 1x2 switches and 80 shutters/VOAs.




                                        Page 116 of 144
                           IST IP NOBEL "Next generation
                                                               D20: First Guidelines for Technologies
                           Optical network for Broadband
                                                                Enabling broadband optical networks
                                European Leadership"
                                                                            ee787c6d-c81c-433e-a3ea-
                                                                                   3ab4c127c833.doc




 Figure 7-38 – Waveguide layout. The bottom diagram shows the 1x2 switch design, in which the
  marked angles represent the accumulated phase difference between the local eigenmodes of a
                                                         xci
                                     particular section.

Reported measurements (Figure 7-39) show a pass-through insertion loss below 5.5 dB,
isolation greater than 45 dB and a low PDL.




                                      Page 117 of 144
                             IST IP NOBEL "Next generation
                                                                    D20: First Guidelines for Technologies
                             Optical network for Broadband
                                                                     Enabling broadband optical networks
                                  European Leadership"
                                                                                  ee787c6d-c81c-433e-a3ea-
                                                                                         3ab4c127c833.doc




Figure 7-39 – Measured transmissivities between the “in” and “drop” ports for various configurations
of sending all channels to the same drop port, starting with port 1 in the upper left plot. Each graph
                                                                             xci
                        shows an overlay of all eight port transmissivities.

On the basis of this building block, further steps towards integration have been
demonstrated, including interleaver/deinterleaver integration, band
multiplexer/demultiplexer and optical monitorsxcii.

  7.7.1.2 WSS 1xN with common and individual channel drop
A ROADM with WSS functionality is represented in Figure 7-40, together with spectrum at
input, output, common add and common drop portsxciii.
The structure (implemented in a compact 95x25 mm die) is similar to a ROADM based on
2x2 switches, but signals can be routed to separate wavelength-specific add/drop ports or
to common-wavelength ports via thermo-optic switches. Insertion loss was about 6 dB, with
more than 40 dB isolation for all channel routing permutations.




                                         Page 118 of 144
                           IST IP NOBEL "Next generation
                                                               D20: First Guidelines for Technologies
                           Optical network for Broadband
                                                                Enabling broadband optical networks
                                European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




                                                                         xciii
                      Figure 7-40 – WSS structure and optical spectrum


  7.7.1.3 WSS 1xN with transversal filter configuration
Another approach for WSS that has been recently demonstrated is based on transversal
filter configurationxciv.
The device (Figure 7-41) is composed by a tree structure with 1 x 4 splitter and by a mesh
structure 4 x 4 combiner composed of 4 couplers. The combination of splitter and combiner
forms a 1 x 4 optical switch based on a transversal filter configuration.
The input signal is split and demultiplexed into 5 wavelengths by the first AWGs; then the
signals experiences a phase shifts via the thermo-optic phase shifters (a proper phase
shifts is required to route a given wavelength channel to four outputs), before being
multiplexed by the second AWG. The four multiplexed components are mixed in the
combiner and are finally output to one of four output ports.
Among the properties of this implementation, there is the constant electrical power required
(the total heat evolution is constant and the output selection does not affect the chip
temperature control system for thermally sensitive AWGs) and that insertion loss does not
depend on the output.




                                      Page 119 of 144
                            IST IP NOBEL "Next generation
                                                                  D20: First Guidelines for Technologies
                            Optical network for Broadband
                                                                   Enabling broadband optical networks
                                 European Leadership"
                                                                               ee787c6d-c81c-433e-a3ea-
                                                                                      3ab4c127c833.doc




   Figure 7-41 – Schematic configuration of 5 wavelength-selective optical 1 x 4 switch based on
                                                                xciv
                                transversal filter configuration .

Figure 7-42 shows examples some wavelength routing conditions. In (a) all the channels
are directed to output 3. In (b) only λ2 is diverted to output 2. In (c), λ1, λ2, λ3, λ4 and λ5
are routed to outputs 4, 3, 2, 3 and 2, respectively. In (d), λ1, λ2, λ3, λ4 and λ5 are routed
to outputs 1, 4, 3, 3 and 2, respectively.
The average insertion loss was 5.7 dB, while a loss deviation between four output ports for
a given wavelength light is less than 0.1 dB (path-independent loss characteristics for each
wavelength).




                                        Page 120 of 144
                            IST IP NOBEL "Next generation
                                                                  D20: First Guidelines for Technologies
                            Optical network for Broadband
                                                                   Enabling broadband optical networks
                                 European Leadership"
                                                                               ee787c6d-c81c-433e-a3ea-
                                                                                      3ab4c127c833.doc




                                                                                           xciv
        Figure 7-42 – Transmission spectra for different wavelength switching conditions


  7.7.1.4 WSS with Free Space Optic
Capella Photonics is promoting the WSS WavePath family (based on Free Space Optic):
Wavepath 9000, The WavePath 9000, designed for long haul and regional applications,
supports up to 90 channels in the C-band spaced at either 50 GHz or 100 GHz. It input,
output and 8 service ports (configurable as add or drop).
The patented Capella Photonics approachxcv (Figure 7-43) is based on a wavelength-
dispersing means such as a diffraction grating to spatially separate a multi-wavelength
optical signal along with a reference signal by wavelength into multiple spectral channels
and a reference spectral component in a spectral array with a predetermined relative
alignment. By aligning the reference spectral component at a predetermined location, the
spectral channels simultaneously impinge onto designated locations, e.g., on an array of
beam-receiving elements positioned in accordance with the spectral array. The reference
spectral component may be further maintained at the predetermined location by way of
servo-control, thereby ensuring that the spectral channels stay aligned at the designated
locations.




                                        Page 121 of 144
                           IST IP NOBEL "Next generation
                                                                D20: First Guidelines for Technologies
                           Optical network for Broadband
                                                                 Enabling broadband optical networks
                                European Leadership"
                                                                            ee787c6d-c81c-433e-a3ea-
                                                                                   3ab4c127c833.doc




                                                                                         xcv
       Figure 7-43 – Schematic of wavelength separating-routing from Capella Photonics


  7.7.1.5 WSS with Free Space optic and 2-axis micromirror
1xN WSS built with MEMS are often limited to a maximum port count of N = 4, limited by
the optical diffractionxcvi.
A method to increase the port count from N to N2 by using a 2D collimator array in
conjunction with two-axis beam-steering functions has been proposed and a 1x2 WSS with
50-GHz channel spacing was experimentally demonstrated xcvii.
As depicted in Figure 7-44 the WDM signal is spatially separated by the diffraction grating
and focused onto its corresponding micromirror by the resolution lens. Then the 2-axis
mirrors direct individual wavelengths to any arbitrary output port in the 2D collimator array
(used in conjunction with a telescope beam expander to reduce the optical spot size on the
MEMS mirror).




                                      Page 122 of 144
                             IST IP NOBEL "Next generation
                                                                    D20: First Guidelines for Technologies
                             Optical network for Broadband
                                                                     Enabling broadband optical networks
                                  European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




Figure 7-44 – Schematic of the 1xN2 wavelength-selective switch using a 2-axis analog micromirror
                                                                                                xcvii
 array and a monolithic 2D collimator array. The telescope expands the size of the optical beam

The devices, fabricated using the SUMMiT-V surface micromachining process provided by
Sandia National Laboratory, are represented in (scanning electron micrograph of the 2-axis
micromirror) and (SEM picture of the array).




                                                                                                  xcvii
   Figure 7-45 – SEM picture of the 2-axis micromirror with the left half of the mirror removed




                                         Page 123 of 144
                          IST IP NOBEL "Next generation
                                                                D20: First Guidelines for Technologies
                          Optical network for Broadband
                                                                 Enabling broadband optical networks
                               European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




                                                                               xcvii
                Figure 7-46 – SEM pictures of the two-axismicromirror array.

Basing on this approach, a 1x32 WSS has been demonstratedxcviii, using a large scan-
angle in both axes (+/- 6.7 degrees mechanical), two-axis analog micromirror array. The
optical insertion loss is less than 5.67 dB, with extinction ratio of 30 dB and 0.5 msec
switching time. The channel spacing is 100 GHz.



  7.7.1.6 WSS with hybrid integration of planar waveguides and MEMS
Metconnex implements Wavelength Selective Switching through a patent-pending hybrid
integration of planar waveguides and MEMS (Micro Electro-Mechanical Systems), with a
product named WSS 5400xcix.
As shown in Figure 7-47, the waveguides provide the demultiplexing and multiplexing
function, while the MEMS provides the any to any switching capability. Their integration is
used to provide wavelength blocking as well as per wavelength and/or per port attenuation.




                                      Page 124 of 144
                          IST IP NOBEL "Next generation
                                                             D20: First Guidelines for Technologies
                          Optical network for Broadband
                                                              Enabling broadband optical networks
                               European Leadership"
                                                                          ee787c6d-c81c-433e-a3ea-
                                                                                 3ab4c127c833.doc




                                                                   xcix
                         Figure 7-47 – Block diagram of WSS 5400

The WSS 5400 is a 10 port, 100 GHz device that directs each input wavelength to any one
of nine output ports, also managing per-wavelength and per-port attenuation.
Design characteristics and technology of this hybrid approach has been discussedc,
focusing of WSS 1x9 cascabeability, thanks to wide flat passbands, low loss and
dispersion.
MEMS are used for switching and waveguide for filtering, but the waveguide filter is
optimized for use with a free-space interconnection. The L-shaped elements in Figure 7-48
are waveguide diffraction gratings emitting an anamorphic beam at the waveguide facet
dispersed continuously with a wavelength dependent angle. The wavelength channel
beams are collimated and then focused onto respective MEMS elements, that are able to
tilting in two dimensions, to establish independent lightpaths for each wavelength channel
between the incoming demultiplexing dispersive element and any of the output multiplexing
dispersive elements. The mirrors are controlled in an analog fashion as to enable VOA by
misalignment of the beams.
Over 39 channels at 100GHz spacing, the passband at 0.5dB is >50GHz. The isolation is
typically >35dB over +/-12.5GHz. The cascaded filtering function does not add significant
degradation to the signal up to 32 cascades (the additional penalty remains below 0.6 dB
for all measured channels).




                                     Page 125 of 144
                             IST IP NOBEL "Next generation
                                                                   D20: First Guidelines for Technologies
                             Optical network for Broadband
                                                                    Enabling broadband optical networks
                                  European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




  Figure 7-48 – The device incorporates a 2-layer stack of arrays of waveguide diffraction gratings
                                                                               c
                   interconnected through free-space to a MEMS mirror array.




 7.8     Other structures/technologies for ROADM applications
The categories discussed above allow performing first level considerations on ROADM
performances, while some specific proposed technologies might not completely fit these
categories.
This section presents some alternative solutions for ROADM.

 7.8.1      FWM in SOA
A possible solution to operate a tunable OADM is to exploit FWM effect in non linear
devices, such as SOA: this approach has been demonstratedci and the concept scheme is
depicted in Figure 7-49.
The optical signal at the input is split in two components, corresponding to pass-through
and drop signal, each one coupled with a different control signal (control laser with timing
information recovered from the input signal), and then entering a SOA.
In each nonlinear device one of the optical control clocks with a carrier frequency f2, is
coupled, which differs about 1 THz from the signal carrier frequency f1. To let a signal
pulse pass an optical path, the control pulse is fed into the corresponding nonlinear device
simultaneously with the signal pulse, which results in an additional pulse at the four-wave
mixing carrier frequency f3 = 2f2 - f1. A band-pass filter (BP) is used to select f3 passing
the optical path, whereas the other frequencies f1 and f2 are suppressed.
With this scheme, properly managing the activation of the control laser pulse it is possible
to achieve flexible switching (in principle it is possible to provide any arbitrary control pulse
sequence so that any data bit sequence can be coupled to any output).
Together with a capable control mechanism, this scheme permits burst or even packet
switching in high-speed optical networks.




                                         Page 126 of 144
                           IST IP NOBEL "Next generation
                                                               D20: First Guidelines for Technologies
                           Optical network for Broadband
                                                                Enabling broadband optical networks
                                European Leadership"
                                                                              ee787c6d-c81c-433e-a3ea-
                                                                                     3ab4c127c833.doc




                                                                         ci
                   Figure 7-49 – Concept of an ADM based on FWM in SOA .




7.8.2 Photonic Crystals
One promising technology is based on Photonic Crystals; achievements till now are
relevant to configurable OADMs, but it is worth considering it due to high potential
improvements.
As an example, a configurable OADM based on Bragg reflectors and coupled waveguides
(Figure 7-50) has been implementedcii basing on autocloned Photonic Crystals.




                                                                                     cii
           Figure 7-50 – Schematic illustration of an OADM based on autocloned PCs

With autocloning (a manufacturing approach jointly developed by Tohoku University -
Sendai, Japan and Photonic Lattice Inc) a corrugated stack resembling cardboard is
assembled from two materials with different refractive indexes; in this example, a photonic
waveguide and the Bragg reflectors were built with alternating corrugated layers made from
a tantalum-oxygen compound (Ta2O5) and silicon dioxide (Figure 7-51). The core
waveguide region was formed from straight alternating layers, and the cladding region on
both sides that confine the photons had the characteristic wave configuration. The
manufacturing challenge is to get an accurate and reproducible pattern at 500-nanometer
dimensions. The small dimensions are critical, since the variation in refractive index has to
be on the order of one-half the wavelength of the light with which the material is interacting.
Transmission loss was 0.56 dB/mm and the fiber coupling is low-loss (the waveguide
structure shapes the optical beam into a round configuration).




                                      Page 127 of 144
                             IST IP NOBEL "Next generation
                                                                    D20: First Guidelines for Technologies
                             Optical network for Broadband
                                                                     Enabling broadband optical networks
                                  European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




    Figure 7-51 – Schematic illustration of a 3D PC structure that operates as a Bragg reflector.

Figure 7-52 shows the measured output optical spectra from port 2 (drop-out; solid line)
and from port 4 (through-out; dashed line). At the Bragg wavelength (1560 nm), the output
power from port 4 was decreased by 16 dB, most of which was coupled to port 2 (loss at
1557 nm are due to radiation loss and can be avoided changing refractive indexes at the
top and bottom cladding layers). The drop bandwidth at FWHM was about 2.0 nm.




      Figure 7-52 – Measured optical spectra for port 2 (drop-out) and for port 4 (through-out)




                                         Page 128 of 144
                            IST IP NOBEL "Next generation
                                                            D20: First Guidelines for Technologies
                            Optical network for Broadband
                                                             Enabling broadband optical networks
                                 European Leadership"
                                                                        ee787c6d-c81c-433e-a3ea-
                                                                               3ab4c127c833.doc




 8       Raman amplification
Introduction and principles of Raman effect.


 8.1     Raman Gain and Raman Amplification
Gain in different fiber types
Gain shaping
Raman amplifier configuration and LRA (Lumped Raman Amplifiers)


 8.2     Noise Components and Non-linear effects
Souces of noise in Raman amplifiers
Noise dependencies from fiber/gain/configuration/…
Impact of non-linearities



 8.3     LRA in Transmission Systems and comparison with EDFA
Comparison LRA/EDFA
Cascading LRA (ref. to WP5)
C-band, L-band,…Wideband Raman Amplifiers



 8.4     LRA Products
Overview of products and special fibers for LRA




                                      Page 129 of 144
             IST IP NOBEL "Next generation
                                             D20: First Guidelines for Technologies
             Optical network for Broadband
                                              Enabling broadband optical networks
                  European Leadership"
                                                         ee787c6d-c81c-433e-a3ea-
                                                                3ab4c127c833.doc




9   Behavior of amps and amp chains




                       Page 130 of 144
              IST IP NOBEL "Next generation
                                              D20: First Guidelines for Technologies
              Optical network for Broadband
                                               Enabling broadband optical networks
                   European Leadership"
                                                          ee787c6d-c81c-433e-a3ea-
                                                                 3ab4c127c833.doc




10 Adaptive filters PMD, fast             -filters




                        Page 131 of 144
                     IST IP NOBEL "Next generation
                                                     D20: First Guidelines for Technologies
                     Optical network for Broadband
                                                      Enabling broadband optical networks
                          European Leadership"
                                                                 ee787c6d-c81c-433e-a3ea-
                                                                        3ab4c127c833.doc




11 Tunable filters

11.1 MEMS-based tunable filters

11.1.1   AXSUN Fabry-Perot MEMS tunable filter

11.1.2   IOLON Diffraction Grating MEMS tunable filters

11.2 Compliant MEMS-based tunable filters

11.2.1   SOLUS MICRO C-MEMS tunable filter

11.3 Tunable Thin Film Filters

11.3.1   AEGIS TFF tunable filter

11.3.2   MICRONOPTICS TFF tunable filter

11.3.3   SANTEC TFF tunable filter

11.4 Liquid Crystal tunable filters

11.4.1   DIGILENS LC tunable filter

11.4.2   NANOOPTO LC tunable filter

11.4.3   LC as cladding of SOI waveguides

11.4.4   Liquid Crystal in combination with Photonic Crystals and Photonic
         Crystal Fibers

11.5 Tunable Fiber Bragg Grating (FBG)

11.5.1   AOS tunable FBG

11.5.2   FBG Strain tuning




                               Page 132 of 144
                      IST IP NOBEL "Next generation
                                                      D20: First Guidelines for Technologies
                      Optical network for Broadband
                                                       Enabling broadband optical networks
                           European Leadership"
                                                                  ee787c6d-c81c-433e-a3ea-
                                                                         3ab4c127c833.doc




11.5.3   ALNAIR tunable FBG

11.5.4   PLC Bragg Grating

11.6 Volume Bragg Grating (VBG) tunable filters

11.7 PLC with micro-resonators tunable filters


11.7.1   Tunable Drop filter with micro-rings in HIC materials

11.7.2   Polymer micro-rings

 11.7.2.1 LITTLE OPTICS micro-ring based tunable filter

11.8 Acousto-optic tunable filters (AOTF)

11.8.1   Etched cladding AOTF

11.8.2   Acousto-optic polarimeter

11.9 Other tunable filter technologies and commercial devices

11.9.1   JDSU tunable filter

11.9.2   OPTUNE tunable filter

11.9.3   DICON tunable filter

11.9.4   All-Fiber tunable devices

11.9.5   Sampled Grating in InGaAsP/InP deep-ridge waveguide




                                Page 133 of 144
           IST IP NOBEL "Next generation
                                           D20: First Guidelines for Technologies
           Optical network for Broadband
                                            Enabling broadband optical networks
                European Leadership"
                                                       ee787c6d-c81c-433e-a3ea-
                                                              3ab4c127c833.doc




12 Compensators/FEC




                     Page 134 of 144
            IST IP NOBEL "Next generation
                                            D20: First Guidelines for Technologies
            Optical network for Broadband
                                             Enabling broadband optical networks
                 European Leadership"
                                                        ee787c6d-c81c-433e-a3ea-
                                                               3ab4c127c833.doc




13 Wavelength–Conversion




                      Page 135 of 144
            IST IP NOBEL "Next generation
                                            D20: First Guidelines for Technologies
            Optical network for Broadband
                                             Enabling broadband optical networks
                 European Leadership"
                                                        ee787c6d-c81c-433e-a3ea-
                                                               3ab4c127c833.doc




14 Regeneration: 2R/3R




                      Page 136 of 144
             IST IP NOBEL "Next generation
                                             D20: First Guidelines for Technologies
             Optical network for Broadband
                                              Enabling broadband optical networks
                  European Leadership"
                                                         ee787c6d-c81c-433e-a3ea-
                                                                3ab4c127c833.doc




15 Optical switching




                       Page 137 of 144
             IST IP NOBEL "Next generation
                                             D20: First Guidelines for Technologies
             Optical network for Broadband
                                              Enabling broadband optical networks
                  European Leadership"
                                                         ee787c6d-c81c-433e-a3ea-
                                                                3ab4c127c833.doc




16 OPM and signal quality analysis




                       Page 138 of 144
                               IST IP NOBEL "Next generation
                                                                     D20: First Guidelines for Technologies
                               Optical network for Broadband
                                                                      Enabling broadband optical networks
                                    European Leadership"
                                                                                   ee787c6d-c81c-433e-a3ea-
                                                                                          3ab4c127c833.doc




    17 Summary




    18 References


1
 L. Paraschis, O. Gerstel, R. Ramaswami, “Evaluation of Tunable Laser Applications in
Metropolitan Networks”, OFC 2004 MF105
2
    M.-C. Amann and J. Buus,” Tunable Laser Diodes” Artech House, 1998.
3
  B. Pezeshki, E. Vail, J. Kubicky, G. Yoffe, S. Zou, J. Heanue, P. Epp, S. Rishton, D. Ton, B. Faraji,
M. Emanuel, X. Hong, M. Sherback, V. Agrawal, C. Chipman, and T. Razazan , “20mW widely
tunable laser module using DFB array and MEMs selection”, OFC 2002
4
 B. Pezeshki et al., “12 element multi-wavelength DFB arrays for widely tunable laser modules”,
OFC 2002 ThGG71
5
 J. Heanue, “Widely Tunable Laser Module Using DFB Array and MEMs Selection with Internal
Wavelength Locker”, OFC2003 MF67
6
    T. Munks et al., “Integrated packaging advances tunable lasers”, Lightwave February 2004
7
 Santur, “TL-2010, TL-2020-C High-Power Widely Tunable Laser Module”, Datasheet March 2003
DCN-00200-OPTL-2010/20-C Ver. 1 www.marubun.co.jp/comnet
8
 Paoletti R. et al, “Small Chip Size, Low Power Consumption, Fully Electronic Controlled Tunable
Laser Source with 40 nm Tuning Range and 20 mW Output Power for WDM Applications”, ECOC
2003 Th1.2.4
9
 H. Hatakeyama, K. Naniwae, “Wavelength-selectable microarray light sources for S-, C-, and L-
bands WDM applications”, OFC 2002 WF2
10
  Y. Tohmori, “Wavelength-selectable DBR Laser Array Inherently Free from Mode-hopping for
High-speed Switching”, OFC2003 ThF5
11
 QDI, “WSL Tunable DFB Laser Array Module With Wavelength Locker”, Datasheet
www.qditechnology.com/
12
   B. Puttnam, M. Düser, P. Bayvel, T. Mullane, T. Farrell, D. McDonald, “Experimental Investigation
of Rapid Wavelength-Switching (<80 ns) in Fast Tuneable Lasers for Applications in Optical Packet
and Burst-Switched Networks”, London Communications Symposium 2003




                                          Page 139 of 144
                              IST IP NOBEL "Next generation
                                                                   D20: First Guidelines for Technologies
                              Optical network for Broadband
                                                                    Enabling broadband optical networks
                                   European Leadership"
                                                                                ee787c6d-c81c-433e-a3ea-
                                                                                       3ab4c127c833.doc




13
  J. Simsarian, A. Bhardwaj, J. Gripp, K Sherman, Y. Su, C.Webb, L. Zhang, M. Zirngibl, “Fast
Switching Characteristics of a Widely Tunable Laser Transmitter”, IEEE PHOTONICS
TECHNOLOGY LETTERS, VOL. 15, NO. 8, AUGUST 2003
14
  M. C. Larson et al., “High Performance Widely-Tunable SG-DBR Lasers”, SPIE Photonics West,
2003, Paper 4995-13
15
   F. Delorme, B. Pierre, H. de Bailliencourt, G. Terol and P. Devoldere, “Wavelength Aging Analysis
of DBR Lasers Using Tuning Section IM Frequency Response Measurements”, IEEE Photon.
Technol. Lett., vol. 11, pp. 310 –312, 1999.
16
  C. Coldren, “Reliability of Widely-Tunable SGDBR Lasers Suitable for Deployment in Agile
Networks”, OFC2003 MF60
17
  Agility, “Agility 3205/3206 CW Widely Tunable Laser Assembly”, Preliminary Product Data Sheet
June 10, 2004 MKS-052-C, www.agility.com
18
 M.L.Majewski, J. Barton and L.A. Coldren, Y. Akulova and M.C. Larson, “Widely tunable directly
modulated sampled-grating DBR lasers”, OFC 2002 ThV2 2
19
     Coldren, “Integrated Tunable Transmitters for WDM Networks”, ECOC 2003 Th1.2.1
20
  Y.A. Akulova, C. Schow, A. Karim, S. Nakagawa, P. Kozodoy, G.A. Fish, J. DeFranco, A. Dahl, M.
Larson, T.Wipiejewski, D. Pavinski, T. Butrie, and L.A. Coldren, “Widely-Tunable Electroabsorption-
Modulated Sampled Grating DBR Laser Integrated with Semiconductor Optical Amplifier”, OFC
2002 ThV1
21
  Y. A. Akulova, G. A. Fish, P. Koh, P. Kozodoy, M. Larson, C. Schow, E. Hall, H. Marchand, P.
Abraham, L. A. Coldren, “10 Gb/s Mach-Zehnder modulator integrated with widely-tunable sampled
grating DBR Laser”, OFC 2004 TuE4
22
 S. Nakagawa, G. Fish, A. Dahl, P. Koh, C. Schow, M. Mack, Wang, R. Yu, “Phase Noise of
Widely-Tunable SG-DBR Laser”, OFC 2003 ThF2
23
  D J Robbins, G Busico, L Ponnampalam, J P Duck, P J Williams, R A Griffin, A J Ward, D C J
Reid, N D Whitbread, E Barton, B Reid and K Kasunic, “A high power, broadband tuneable laser
module based on a DS-DBR laser with integrated SOA”, OFC 2004 TuE3
24
  Robbins D.J., Duck J.P., Ponnampalam L., Busico G., Griffin R.A., Reid D.C.J., Williams P.J.,
Ward A.J., Whitbread N.D., Carter A.C. “Co-Packaging of Monolithic, Full C-Band Tuneable Lasers
with GaAs Modulators for 10Gbps Transmitters”, ECOC 2003 Th1.2.2
25
   F. Kano, Y. Yoshikuni, M. Wakamiya, A. Kanagawa, H. Ishii, “Frequency control and stabilization
of broadly tunable SSG-DBR lasers” OFC 2004 ThV3
26
  Y. Gustafsson, J. Wesström and P. Szabo, “On the Frequency Stability of Widely Tunable GSCR
Lasers”, OFC2002 ThGG82
27
  O. Wesström, G. Sarlet, S. Hammerfeldt, L. Lundqvist, P. Szabo, P.-J. Rigole, "State-of-the-art
performance of widely tunable modulated grating Y-branch lasers", OFC 2004 TuE2
28
 Syntune, “S1500 - Widely tunable laser module with wavelength locker”, Product Information,
www.syntune.com
29
   Intune, “AltoNet1200TM Fast Wavelength-Switched Tunable Laser Module”, Product Datasheet
altonet1200-0903-a, www.intune-technologies.com
30
  D. Anthon, J. D. Berger, and A. Tselikov, “C+L band MEMS tunable external cavity semiconductor
laser”, OFC2004 WL2
31
 D. Anthon, J. Berger, K. Cheung, A. Fennema, S. Hrinya, H. Lee, A. Tselikov, “Frequency and
Mode Control of Tunable External Cavity Semiconductor Lasers”, OFC 2003 MF61




                                         Page 140 of 144
                               IST IP NOBEL "Next generation
                                                                    D20: First Guidelines for Technologies
                               Optical network for Broadband
                                                                     Enabling broadband optical networks
                                    European Leadership"
                                                                                 ee787c6d-c81c-433e-a3ea-
                                                                                        3ab4c127c833.doc




32
     Iolon, “Apollo™ Widely Tunable Laser”, Datasheet 20-01549 Revision B, www.iolon.com
33
     Intel, “Performance and Design” White Paper May 2003, www.intel.com
34
  M. Finot et al., “Automated Optical Packaging Technology for 10 Gb/s Transceivers and its
Application to a Low-Cost Full C-Band Tunable Transmitter”, Intel® Technology Journal, Volume 08
Issue 02 Published, May 10, 2004
35
     Intel, “TTX11500 Full C-Band Tunable Laser”, Product Brief, www.intel.com
36
  Princeton Optronics, “High-Power Tunable Laser - Release 2.0, LC242000-01”, Product Data
Sheet February 4, 2004, www.i-waveco.com
37
  Bergonzo A., Brenot R., Locatelli D., Landreau J., Becaert Vuong A., Lafragette J.L., Durand O.,
Jacquet J., “12-Channels 200GHz-Spacing Digitally Tunable External Cavity Laser with Fast
Wavelength Switching”, ECOC 2003 Th1.2.5
38
  A.Bergonzo et al., “A Very Simple and Efficient Widely Tunable Sampled Fiber Bragg Grating
External Cavity Laser” OFC 2003 ThF3
39
 Pirelli, “DTL C13 Series - High Power Full C-Band tunable laser source“, Product Datasheet,
www.pirelli.com
40
  Sven Mahnkopf, Martin Kamp, Martin Arlt, Reinhard März, Francois Lelarge, Guang-Hua Duan,
and Alfred Forchel, “Widely Tunable Complex-Coupled Distributed Feedback Laser With Photonic
Crystal Mirrors and Integrated Optical Amplifier”, IEEE PHOTONICS TECHNOLOGY LETTERS,
VOL. 16, NO. 3, MARCH 2004 729
41
  Kazumasa Takabayashi, Kan Takada, Naoki Hashimoto, Masaharu Doi, Syuichi Tomabechi,
Tatsuya Takeuchi, Goji Nakagawa, Hideyuki Miyata, Tadao Nakazawa and Ken Morito, “Widely (90
nm) Wavelength Tunable Laser Using a Semiconductor Optical Amplifier and an Acousto-Optic
Tunable Filter”, ECOC 2003 Th1.2.6
42
  S. Matsuo et al., “Digitally Tunable Laser Using Ladder Filter and Ring Resonator”, ECOC 2003
Th1.2.3
43
  S. Matsuo, S.-H. Jeong, T. Segawa, H. Okamoto, Y. Kawaguchi, Y. Kondo, H. Suzuki, Y.
Yoshikuni, “Monolithically Integrated Digitally Tunable Laser Using Ladder Filter and Ring
Resonator”, ECOC 2004 We4.P.062
44
   C. C. Renaud, M. Dueser, B. Puttnam, T. Lovell , P. Bayvel, A. J. Seeds, “Nanosecond switching
time, uncooled, zero frequency error DWDM source”, OFC 2004 WL3
45
  H. Yamazaki, M. Ishizaka, M. Takahashi, S. Watanabe, K. Suzuki, K. Sato, S. Sudo, “Widely
Tunable Laser Consisting of a Silica Waveguide Double Ring Resonator Connected Directly to a
Semiconductor Optical Amplifier”, ECOC 2004 Th4.2.4
46
 M. Feuer and D. Al.Salameh, “Routing power: a metric for reconfigurable wavelength Add/Drop
modules”, Proceedings OFC 2002, Paper TuX1
47
  C.R. Giles and M. Spector, “The wavelength add/drop multiplexer for lightwave communication
networks”, Bell Labs technical Journal, vol. 4, pp. 207-229 (1999)
48                                                                                 nd
  R. Ramaswami, K. N. Sivarajan, “Optical Networks – A Practical Perspective”, 2        Ed., Morgan
Kaufmann Publ., October 2001.
49
  J. Ford, “Micromechanical Wavelength Add/Drop Switching: From Device to Network
Architecture”, Proceeding OFC 2003, Paper TuN3

50
 V. Viscardi, G. Barozzi and O. Gerstel, “Asymmetric Reconfigurable OADMs for next generation
Metro-DWDM networks”, TuH3 OFC2004




                                          Page 141 of 144
                              IST IP NOBEL "Next generation
                                                                    D20: First Guidelines for Technologies
                              Optical network for Broadband
                                                                     Enabling broadband optical networks
                                   European Leadership"
                                                                                  ee787c6d-c81c-433e-a3ea-
                                                                                         3ab4c127c833.doc




51
  Santec, “Configurable OADM module - iOADM”, Product Datasheet i-OADM-C-E / Ver.2.1 CODE-
200303-NS-AI-CPY March 4, 2003 www.santec.com
52
     Optun, “A-block module”, Product Datasheet V23845-Z1199-A0001, www.optun.com
53
     Dupont, “iROAD888”, Product Datasheet, www.photonics.dupont.com
54
     Global Opticom, “Programmable Add Drop Multiplexer”, Datasheet, www.globalopticom.com
55
     Chromux, “Roadplexer”, Spec Sheet Version 2.0, www.chromux.com
56
     Hoya, “RC-OADM”, Preliminary data, www.hoya.com
57
  T. Kanie, “Ultra-Compact Multichannel Optical Components Based on PLC and MEMS
Technologies”, OFC2003 WM4
58
     C.R. Doerr, “Optical Add/Drops and Cross Connects in Planar Waveguides”, Th1.3 ECOC 2003
59
  A. Schumacher, “Fully Reconfigurable, 20 Channel Optical Add-/Drop Multiplexer with Integrated
Variable Optical Attenuators and Power Monitors”, TuE3 OFC2003
60
  T. Spott “Macroscopic and Microscopic Thermal Management of a Planar Integrated Optical
Device”, TuE6 OFC2003
61
  L. Eldada, J. Fujita, R. Gerhardt, A. Radojevic and O. Zhuromskyy, “Channel
Switching/Monitoring/Equalizing/Shuffling Subsystem on a chip for Reconfigurable OADM”, IPR
2003
62
 A. M. Radojevic, J. Fujita, R. U. Gerhardt, T. Izuhara, and L. A. Eldada, “Hybrid-Integrated
ROADM for Reconfigurable Optical Networks”, TuI4.5 ECOC 2004
63
  N. Keil, H. Yao, C. Zawadzki, O. Radmer, F. Beyer, M. Bauer, C. Dreyer, J. Schneider, “Recent
Progress in Polymer based Photonic Devices”, ECOC 2004 THI4.1
64
     R. B. Sargent, “Recent Advances in Thin Film Filters”, OFC2004 TuD6
65
   T. Ito, Y. Iida, K. Minagawa, K. Aihara, K. Mizuno, Y. Mimura, “Optimization for CD Characteristic
of Band-OADM Using 4-skip-0 100GHz Filter”, OFC 2003 ThM2
66
  C. Dragone, “Planar waveguide devices for efficient switching and wavelength routing”, OFC 2004
TuI1
67
     Bookham, “Flexible Upgrade Architecture”, Solution Presentation www.bookham.com
68
     Lynx, “Photon.8x8 PCSS”, Datasheet MK/1203/R1 www.lynx.com
69
     Dupont, “iROAD800 8x8 optical switch”, Product Datasheet , www.photonics.dupont.com
70
 W. Chen, “Monolithically Integrated 32-channel Client Re-configurable Optical Add/Drop
Multiplexer on Planar Lightwave Circuit”, OFC2003 TuE5
71
 Z. Zhu, W. Chen, Y. Chen, J. Sun, D. Huang, J. Chen, “Cascade-able Client-reconfigurable
Optical Add-Drop Multiplexer”, ECOC 2004 TuI5.1
72
 M. Okuno, “Highly Integrated PLC-Type Optical Switches for OADM and OXC Systems”,
OFC2003 TuE1
73
 Y. Hashizume, Y. Inoue, T. Kominato, T. Shibata, and M. Okuno, “Low-PDL 16-channel Variable
Optical Attenuator Array using Silica-based PLC”, OFC 2004 WC4
74
     Dupont Photonics Technologies, “iOXCTM 800”, Datasheet www.photonics.dupont.com
75
  Santec, “Lambda Blocker”, Preliminary Datasheet Ver.1.0 CODE-200208-TM-AI-CPY August
2002




                                         Page 142 of 144
                                 IST IP NOBEL "Next generation
                                                                       D20: First Guidelines for Technologies
                                 Optical network for Broadband
                                                                        Enabling broadband optical networks
                                      European Leadership"
                                                                                     ee787c6d-c81c-433e-a3ea-
                                                                                            3ab4c127c833.doc




76
  JDSU, “Dynamically Reconfigurable Wavelength Blocker for C or Extended L Band”, Product
Bulletin10140148 Rev. 001 03/02
77
       Avanex, “PowerBlockerTM Dynamic Wavelength Blocker”, Datasheet D002SR02/04
78
       LightConnect, “Dynamic Channel Equaliser”, Datasheet MK-PDS-00005-1-0202
79
  Polychromix, ”P-DCO”, 10050AB Dynamic per-channel attenuator and blocker”, Datasheet Ver.
1.6
80
       Xtellus. “Dynamic Block Equalizer”, Preliminary information www.xtellus.com
81
       LightConnect, “Diffractive MEMS”, Whitepaper www.lightconnect.com
82
       S. Senturia , “Dynamic MEMS”, OFC2003 WM3
83
  J. I. Trisnadi, C. B. Carlisle and R. Monteverde, “Overview and applications of Grating Light
Valve™ based optical write engines for high-speed digital imaging”, Photonics West 2004 -
Micromachining and Microfabrication Symposium January 26, 2004
84
       Xtellus, Technology, www.xtellus.com
lxxxv
  Lambda Crossing, “T-OADM – Tunable Add/Drop Multiplexer LambdaFlow™”, Datasheet
www.lambdax.com
lxxxvi
          Optoplex, “Tunable Optical Add/Drop Multiplexer (T -OADM)”, Datasheet www.optoplex.com
lxxxvii
   Ondax, “Single Channel Filter SwitchTM”, Preliminary Datasheet 112-80040-001 Rev A
www.ondax.com
lxxxviii
           AFOP, “Single Channel Optical Add/Drop Module”, Datasheet REV.D 7/30/03 www.afop.com
lxxxix
    C. F. Lam, M. Boroditsky, M.D. Feuer, N. J. Frigo, B. Desai, and J. Kang, “Programmable
Optical Multicasting in a Regional/Metro Area Network Using A Wavelength Selective Optical Cross-
connect”, ECOC 2001
xc
      C. R. Doerr, “Optical Add/Drops and Cross Connects in Planar Waveguides”, ECOC 2003 Th1.3
xci
  C. R. Doerr, L. W. Stulz, D. S. Levy, L. Gomez, M. Cappuzzo, J. Bailey, R. Long, A. Wong-Foy, E.
Laskowski, E. Chen, S. Patel, and T. Murphy, “Eight-Wavelength Add–Drop Filter with True
Reconfigurability”, IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 15, NO. 1, JANUARY 2003
pag.138
xcii
        C. R. Doerr et al., OFC2003 PD
xciii
  J.B.D.Soole, N.Chand, M.P.Earnshaw ∗ , K. Kojima, R. Pafchek, M. Ling, and V. Swaminathan,
“Multipurpose Reconfigurable Optical Add-Drop Multiplexer (ROADM)”, ECOC 2002 PD3.3
xciv
  K. Suzuki, T. Mizuno, M. Oguma, T. Shibata, H. Takahashi, A. Himeno, “Fully-reconfigurable
wavelength-selective optical 1 x N switch based on transversal filter configuration”, WC2 OFC2004
xcv
  Chen et al., “Reconfigurable Optical Add-Drop Multiplexers with servo control and dynamic
spectral power management capabilities”, United Sates Patent No. 6,687,431 B2 Feb. 3, 2004
xcvi
   D. M. Marom, et al., “Wavelength-selective 1x4 switch for 128 WDM channels at 50GHz
spacing,” Proceedings of 2002 OFC postdeadline paper, FB7
xcvii                                     2
   Jui-che Tsai and Ming C. Wu, “1xN Wavelength-Selective Switches with High Fill-Factor Two-
Axis Analog Micromirror Arrays”, OFC 2004 MF42
xcviii
   J. Tsai, L. Fan, C. Chi, D. Hah, M. C. Wu, “A Large Port-Count 1x32 Wavelength-Selective
Switch Using a Large Scan-Angle, High Fill-Factor, Two-Axis Analog Micromirror Array”, ECOC
2004 TuI5.2
xcix
        METCONNEX, “WSS 5400 Product Brief”, www.metconnex.com



                                              Page 143 of 144
                             IST IP NOBEL "Next generation
                                                                   D20: First Guidelines for Technologies
                             Optical network for Broadband
                                                                    Enabling broadband optical networks
                                  European Leadership"
                                                                                ee787c6d-c81c-433e-a3ea-
                                                                                       3ab4c127c833.doc




c
 T. Ducellier, A. Hnatiw, M. Mala, S. Shaw, A. Mank, D. Touahri, D. McMullin, T. Zami, B. Lavigne,
P. Peloso, O. Leclerc, “Novel High Performance Hybrid Waveguide-MEMS 1x9 Wavelength
Selective Switch in a 32-Cascade Loop Experiment”, ECOC 2004 Th4.2.2
ci
 Rohde H., Schairer W., Lehmann G., “All-Optical Add/Drop Multiplexer for High Speed Optical
Networks Based on Four-Wave Mixing (FWM) in SOA”, ECOC 2003 We4.P.127
cii
 M. Shirane, A. Gomyo, K. Miura, Y. Ohtera, H. Yamada, and S. Kawakami, “Optical Add/Drop
Multiplexers Based on Autocloned Photonic Crystals”, OFC 2004 ThR3




                                        Page 144 of 144

								
To top