PowerPoint Presentation

Document Sample
PowerPoint Presentation Powered By Docstoc
					A Parallel Algorithm for Numerical Simulations of
  WDM Optical Fiber Communication Systems
                               Thiab R. Taha
                        Computer Science Department
                           University of Georgia
                            Athens, GA,USA
                     E-mail: thiab@cs.uga.edu
                             ABSTRACT

 Optical fiber communication systems have experienced tremendous
 growth in the last twenty years. In the past year alone researchers
 have announced transmission rates in excess of one tera bit per second.
 At the same time, the growing use of the Internet and the World Wide Web
 has quickly contributed to make this area one of the key technological
 sectors in the global economy, and has generated an unprecedented
 demand for even higher transmission capacities in order to offer a good
 Quality of Service (QoS) .
         A Parallel Algorithm for Numerical Simulation of
           WDM Optical Fiber Communication Systems
                             Thiab R. Taha
                      Computer Science Department
                          University of Georgia
                           Athens, GA,USA
                        E-mail: thiab@cs.uga.edu

                              ABSTRACT
   Optical fiber communication systems have experienced tremendous
   growth in the last twenty years. The recent growth is mainly due to
    the introduction of wavelength division multiplexing(WDM) and
dispersion management(DM) technology. WDM allows the simultaneous
    transmission of multiple optical channels, operating at their own
               frequencies, through the same optical fiber.
In WDM systems, different channel pulses propagate at
different velocities and, as a result, collide with each other. This
leads to signal deteriorations. Another problem for WDM
systems is the presence of resonant four wave mixing (FWM)
terms due to the interaction between the different channels. This
serious problem is partially solved by the introduction of
dispersion management(DM) in optical systems.
In this talk, we introduce a parallel numerical algorithm based
on the spilt step method and the FFT to study the interactions
of WDM dispersion-managed solitons.
Also, other related issues including the polarization effects on
soliton systems will be addressed. The implementation of the
algorithm will be carried out on the SGI origin 2000 parallel
system.
With the current technology, fiber capacity can be as
high as 40 Gbps on a bit stream. As many as 80 bit
streams using WDM can accommodated on a single
glass fiber.

Suppose a fiber optic has 24 strands where we allow
4 spares and 10 in each direction.
The total cable capacity is 40x10x80 = 32,000 Gbps.
 The growing use of the Internet and the World Wide Web
has quickly contributed to make this area one of the key
technological sectors in the global economy, and has
generated an unprecedented demand for even higher
transmission capacities in order to offer a good Quality of
Service (QoS) .
          The new services that are emerging
              in the market are mainly:

Video on Demand, High Speed Internet, Videoconferencing,
         Telemedicine, Gaming, Telelearning, …
Optical Fiber has many advantages over
conventional techniques such as:

1. low loss and attenuation.
2. high bandwidth
3. Its immunity to the electromagnetic interferences.
4. Thin and lightweight, so it is easy to operate.
5. They are more secure against wiretapping.
6. Can extend over longer distances before a repeater is needed.
7. They are more immune to crosstalk within a cable than other
ordinary wires.
Problems with the use of WDM:

1. Due to the periodic distribution of amplifiers, a resonant
instability created by the nonlinear terms (four-wave mixing
(FWM) interactions) can seriously degrade the signal. The
proper use of Dispersion management(DM) can alleviate the
negative effects of FWM.
2. The frequency shifts and the associated displacement in
pulse arrival times created by the interaction of the solitons
with amplifier noise. This can be reduced by the introduction
of guiding filters.
An optical transmission system consists of three
components:

1. the optical transmitter
2. the transmission medium
3. the optical receiver
 The transmitter uses a pulse of light to indicate the ‘1’ bit and
the absence of light to represents the ‘0’ bit. The receiver can
generate an electrical pulse once light is detected.
Two types of fiber:

1. A single mode fiber: this requires the light to
propagate in a straight line along the center of the
fiber. It is used for long distance transmission.
It has a good quality signal.

2. A multimode fiber: a light ray might enter the
fiber at a particular angle and go through the fiber through
internal reflections.
Because of the capacity growth, optical fiber systems
are increasingly being limited by the following
transmission effects:

• Chromatic dispersion,
• Nonlinearity,
• Polarization effects,
• Amplifier noise and others.

In order to design a transmission system, it is crucial to
accurately model and calculate the impairments due to
these effect.
One approach for studying the various physical effects is to
model transmission systems numerically.
Full numerical modeling of real systems is still beyond the
capability of current computational resources in most cases.
As a consequence, there is a critical need for developing
parallel numerical techniques to model these systems.
 In this talk I will discuss some of these techniques.
High Performance Computers
1. In early1980’s computers perform 106 Floating point operations
   per second(Mflop/s)
• Scalar based systems

2. In 1990’s computers perform 109 Floating point operations per
   second(Gflop/s)
• Vector and shared memory computers

 3. Today computers perform 1012 Floating point operations per
    second (Tflop/s)
•Highly Parallel Computers, distributed processing, message passing
4. In 2010 we expect computers to perform 1015 Floating point
   operations per second(Pflop/s)
•Shared/distributed memory processors, many more levels of
 memory hierarchy, more adaptive techniques, extended
 precision.
Top 500 Most powerful Computers in the world
  This list is available from www.top500.org
The National Science Foundation selects Compaq Computer
Corporation and the Pittsburg Supercomputer Center to build
and manage the world largest supercomputer for scientific
applications.

The first delivery of this system is expected by November 2000.
This supercomputer is expected to deliver 5 Teraflops of peak
performance.
 Also, the French Atomic Energy Commission is building the
largest supercomputer in Europe to simulate nuclear testing.

On March 7th. , 2003 Scientists at Stanford University used
fiber-Optic cables to transfer 6.7 Gigabytes of data- the
equivalent of 2 DVD movies- across 6,800 miles in less than
a minute. The data was sent from California to Amsterdam.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:21
posted:9/12/2012
language:English
pages:18