Motivation for OFDM by rgrVhQ3

VIEWS: 37 PAGES: 65

									                                                                                TeNeT Group
                                                                                 IIT Madras

                           Key Elements of
                  Mobile Broadband Access Networks
                                  &
                          Relevance to India


                                      K. Giridhar

                                        Professor
                           Department of Electrical Engineering
                          Indian Institute of Technology Madras
                                     Chennai 600036

                           www.ee.iitm.ac.in; giri@tenet.res.in




                  Devi Ahilya Vishwavidyalaya, Indore, January 27, 2012


K.Giridhar/IITM                       Key Elements in Mobile Broadband Access
                                                                                              1
                                                                               TeNeT Group
                                                                                IIT Madras
Motivation

   Why is Mobile Broadband Access relevant to India?

   What wireless standard offers “true broadband”?
           2.5G, 3G, 3.5G, or 4G?


   Key Elements of Mobile Broadband Access System




         K.Giridhar/IITM             Key Elements in Mobile Broadband Access
                                                                                             2
                                                  TeNeT Group
                                                   IIT Madras
India – Wireless Skills Stack


                                                                Status
                                                                  in
 Value add Innovation Barrier                                   India
 per person  and IP   to Entry

                                     Core                       Nascent
                Highest
                                  Technology

                                   Product
                Medium
                                 Architecture                   Emerging


                                  H/W and S/W
                 Low                                            Strong
                                 Implementation



  KG/Jan.2012
                                                                   3
                                                    TeNeT Group
How can Indian Industry                              IIT Madras

go up the value chain?




Other than applications, protocol-stack implementation, and
various technical services, India should & can

1. Participate in R&D for 4G Wireless Standardisation

                             CEWiT
                               INDIA


2. Foster equipment companies grown ground-up




  KG/Jan.2012
                                                                  4
                                                               TeNeT Group
                                                                IIT Madras
Innovations on the Infrastructure Side



   How can Indian companies enter Mobile Broadband
    (LTE/LTE-A) infrastructure design and/or manufacturing?

   Are there India-specific technical requirements?
       More fixed than mobile users for broadband data
       LoS, near-LoS users, many indoor
       Can voice continue on GSM, or VoIP is essential?

   How significant will be the leveraging of India’s Networking,
    Internet, & Software skills given the all-IP infrastructure?
       How to exploit the fact that most Indian operators have fiber to BS?
       How to exploit Wi-Fi extensions to LTE/LTE-A ?




     KG/Jan.2012
                                                                               5
                                                                          TeNeT Group
Key Advances in Air-Interface                                              IIT Madras


MIMO Precoding
For Capacity or Coverage?
Open loop or Closed loop?
 Single user or Multi user?
                                                  Spatial Multiplexing              Spectral
                        DCS
                                                                                    Efficiency
                                   Interference                 OFDM/OFDMA
                                   ManagementTransmit Diversity
                  Freq. Hopping
                                                           Link
                                          Precoding     Adaptation
                                                         (H-ARQ)
           Multi-user                                                    Sectorisation
           Detection
                                  Conv. Turbo &                 Fixed Beamforming
                                   LDPC Coding
                                                              Receive Diversity
           VAD, VBR
                                                           ST or STF Coding

                  Power Control
                                                             Range, Mobility
 Re-use                  MC-CDMA
Efficiency

    KG/Jan.2009
                                                                                           6
                                                                                        TeNeT Group
                                                                                         IIT Madras
Outline

   What is Co-Channel Interference (CCI)?
           What is the consequence of CCI?
           Methods to mitigate CCI  overview


   IMT-A Standards Specs
           OFDM is the bedrock of all future wireless standards
                  What is OFDM (and its variants)?
                  Why OFDM?
           Overview of IEEE 802.16m and 3GPP LTE


   R&D case studies in CCI mitigation
           Conjugate Data Repetition
           Interference Cancelling Block Modulation
           CCI aware Receiver Processing


   Summary


         K.Giridhar/IITM                      Key Elements in Mobile Broadband Access
                                                                                                      7
                                                               TeNeT Group
                                                                IIT Madras
IIT Madras Campus




           ESB



   K.Giridhar/IITM   Key Elements in Mobile Broadband Access
                                                                             8
                                                                               TeNeT Group
                                                                                IIT Madras
    Foot-print of the Proposed IITM 4G+Test Bed



MEMH                                                                                         ICSR



2x1 GigE optical
      link

4x1 GigE optical
      link



                           Computer Center



                                                                       CSD
         K.Giridhar/IITM             Key Elements in Mobile Broadband Access
                                                                                              9
                                                                          TeNeT Group
Coverage Map in IITM                                                       IIT Madras
(SINR at different locations)




    K.Giridhar/IITM             Key Elements in Mobile Broadband Access
                                                                                        10
                                                              TeNeT Group
                                                               IIT Madras
SINR CDF




  K.Giridhar/IITM   Key Elements in Mobile Broadband Access
                                                                            11
                                                                                          TeNeT Group
                                                                                           IIT Madras
Freq. Reuse, CCI, & Spectral Efficiency


                         Avg: 2 dB       Avg: 14 dB


  pdf of SINR                                                 pdf of SINR
  (reuse 1/1)                                                 (reuse 1/3)

                                                                                             Reuse 1/3


        -10dB                                                                      30dB

                     Avg. SE = 3X b/s (but BW reqd. is 3 times)  X b/s/Hz


                 Avg. SE = X b/s/Hz




                                                                                             Reuse 1/1

   K.Giridhar/IITM                       Key Elements in Mobile Broadband Access
                                                                                                         12
                                                                                  TeNeT Group
                                                                                   IIT Madras
Interference Management/Mitigation -- Overview

           IM at Scheduler
                  Intf. Avoidance
                  Creating “Zones” where Intf. can be mitigated
                     CDR, ICBM


           IM at Tx resource element (subcarrier)
                  Tx Beamforming/SDMA ?
                  Centralised precoding
                  Intf. Alignment
                  Co-operative MIMO

           IM at Rx
                  Synchronisation
                  Channel Estimation
                  Symbol Detection (MIMO Combining)
                  Turbo Decoding




 K.Giridhar/IITM                        Key Elements in Mobile Broadband Access
                                                                                                13
                                                                                        TeNeT Group
                                                                                         IIT Madras
    IM at Scheduler

   Interference Management is possible using BS co-operation

   Exploit IP back-haul to BS; Semi-centralised “upper MAC”/scheduler




                                                                                           Fractional
                                                                                             Reuse



                           A
                           B                                                        C

                                                       A


         Co-operative       GigE back-haul (IP)
          Scheduling


       K.Giridhar/IITM                    Key Elements in Mobile Broadband Access
                                                                                                        14
                                                                                        TeNeT Group
                                                                                         IIT Madras
Outline
   What is Co-Channel Interference (CCI)?
           What is the consequence of CCI?
           Methods to mitigate CCI  overview


   IMT-A Standards Specs
           OFDM is the bedrock of all future wireless standards
                  What is OFDM (and its variants)?
                  Why OFDM?
           Overview of IEEE 802.16m and 3GPP LTE


   R&D case studies in CCI mitigation
           Conjugate Data Repetition
           Interference Cancelling Block Modulation
           CCI aware Receiver Processing


   Summary



         K.Giridhar/IITM                      Key Elements in Mobile Broadband Access
                                                                                                      15
                                                                                    TeNeT Group
                                                                                     IIT Madras
Why OFDM?
   Why OFDM?
        DS-CDMA and TDMA do not scale well with band-width
        Among all block-modulation schemes, OFDM is the most preferred approach
         due to it’s
            (a) simplicity (compared to DS-CDMA)

            (b) flexibility (compared to TDMA)




   UL is either SC-FDMA or OFDMA

                                                Low PAPR +
                                             Narrow-banding                      I-FDMA
                                                Generalised
             Block Mod.




                          SC+CP                 Multi-Carrier
                                                                                 SC-FDMA

                          MC+CP        OFDMA

                          OFDM         For UL
                                  (Narrow-banding)


    K.Giridhar/IITM                    Key Elements in Mobile Broadband Access
                                                                                                  16
                                                                           TeNeT Group
Conventional OFDM (FDE is optional)                                         IIT Madras




Symbol
Mapping
                      Add    Tx                                Conventional
               IFFT   CP
 & S/P                      Mod.                                  OFDM

                                                                                 For Coded OFDM,
                                                                                  LLRs got from CE
                                                                                  (no division by H)

                                                                           CE


                                                                        Frequency
                                Rx            Remove
                                                               FFT        Domain     Detection
                                                CP                                    & P/S
                              Algos.                                     Equaliser




   K.Giridhar/IITM            Key Elements in Mobile Broadband Access
                                                                                              17
                                                                                         TeNeT Group
Single Carrier Trasmission (FDE is mandatory!)                                            IIT Madras




                       Add
    Symbol             CP            Tx                     Tx -- low-complexity, TDMA
    Mapping           (of QAM       Mod.                    Rx -- implements SC-FDE;
                      symbols)
                                                                   Linear Equaliser or DFE


                        to permit FDE                                                        Decision
                                                                                             Feedback
                                                                CE
                                                                                               Filter


                                                          Frequency                         -




                                                                                                  Detector
                      Rx         Sub.      FFT              Domain                IFFT
                    Demod.        CP                       Equaliser


                                            Divide by H                                   For Coded SC system,
                                                                                             LLRs obtained here
                                                                                            (after division by H)
  K.Giridhar/IITM                       Key Elements in Mobile Broadband Access
                                                                                                             18
                                                                               TeNeT Group
MIMO-OFDM                                                                      IIT Madras
Technology Enablers for Mobile Broadband

                              SINR

                                                                Inst. SINR

                                     MCS 2

                                                                Avg. SINR

                                     MCS 1



                                     MCS 0

                                                                   Frequency




       K.Giridhar/IITM               Key Elements in Mobile Broadband Access
                                                                                             19
                                                                                         TeNeT Group
                                                                                          IIT Madras
Wireless Access Standards

                                                                                                  MAN
                                                                                            (non-LoS, mobile,
                                                                                              ped, and fixed)


                               WAN                                                    x Flarion
                            (LoS, fixed)                         x IEEE 802.16d,e     x SOMA
                                                                 x IEEE 802.20        x Arraycomm
                                                                 x HSDPA, HSPA        x Pre-WiMax, Wibro
                                                                 x HDR, EV-DO     x IEEE 802.16m
                                                                 x 3G-DECT        x LTE, LTE-A
                           IEEE 802.16
                                                                                  x IMT-Advanced
                          (10 to 66 GHz)
  Core Network




                                                  base-station

                                                                  1-3km, non-LoS             UE
                            2-10km, LoS
                                                                                       IEEE 802.11a,b,g,n
                                                                                      IEEE 802.15.3 (UWB)
                                                                         LAN            Zigbee, Bluetooth
                      or, Fiber(Ethernet)                             (near LoS,
                                                                      ped, fixed)

        K.Giridhar/IITM                     Key Elements in Mobile Broadband Access
                                                                                                            20
                                                                                        TeNeT Group
                                                                                         IIT Madras
Evolved 3G - Terminology

                           Potential Evolved 3G Architecture

    ‘IMS+’                       Mobility
                                Mobility         HSS/        Session       QoS Policy                 Services
                                Policy           Auth        Control        Server




    ‘SAE’             IP Transport
                        Network
                                                                                            IP Transport
                                                                                              Network
      (System
    Architecture
     Evolution)

                                                IP
                                                                    IP


     ‘LTE’                   RNC

    Long Term                                     WLAN AP
                                                                         E-UMTS AP            Edge Gateway
     Evolution
        or                                                                                    QoS Enabled Router
    ‘E-UTRA’


                                       Courtesy: Dr. Amitava Ghosh, Motorola Labs



    K.Giridhar/IITM                     Key Elements in Mobile Broadband Access
                                                                                                             21
                                                                                                   TeNeT Group
                                                                                                    IIT Madras
802.16m Frame(s)




        Fig illustrates an example TDD frame structure with D:U = 5:3, which is applicable to the nominal channel
        bandwidths of 5, 10, and 20 MHz with G = 1/8. In Figure 465 the last DL AAI subframe, i.e. DL SF4, is a type-3 AAI
        subframe and the other AAI subframes are type-1 AAI subframes. TTG and RTG are 105.714 μs and 60μs,
        respectively. (Courtesy: IEEE 802.16m WMAN)


   K.Giridhar/IITM                               Key Elements in Mobile Broadband Access
                                                                                                                             22
                                                                    TeNeT Group
                                                                     IIT Madras
LTE Frame(s)




                                                              Courtesy: Agilent

  K.Giridhar/IITM
                                  23
                    Key Elements in Mobile Broadband Access
                                                                                  23
                                                               TeNeT Group
                                                                IIT Madras
802.16m OFDM/OFDMA Parameters




   K.Giridhar/IITM   Key Elements in Mobile Broadband Access
                                                                             24
                                                                   TeNeT Group
                                                                    IIT Madras
LTE Resource Structure




   K.Giridhar/IITM
                                       25
                         Key Elements in Mobile Broadband Access
                                                                                 25
                                                                       TeNeT Group
What can MIMO do?                                                       IIT Madras




                     Courtesy: E. Dahlman, IEEE Comm Mag, Apr 2009




   K.Giridhar/IITM           Key Elements in Mobile Broadband Access
                                                                                     26
                                                                               TeNeT Group
                                                                                IIT Madras
Bits to Waveforms – How complex can it get?




          is a codeword (as seen modulation path fed to the MIMO encoder as an input.
 A layer is defined as a coding and by UE)

           is what goes output of the MIMO encoder which is passed to the precoder.
 A stream is defined as an into a Tx antenna


     K.Giridhar/IITM                 Key Elements in Mobile Broadband Access
                                                                                             27
                                                                                    TeNeT Group
Example: DL-MIMO in .16m                                                             IIT Madras




All MIMO modes supported for 2,4,8 Ants.
• Modes 0,1,2 support for single layer only
• Mode 1 and 2 support upto 8-streams
• Modes 3 and 4 support 2,3,4 layers

• LTE   has similar choices


     K.Giridhar/IITM                      Key Elements in Mobile Broadband Access
                                                                                                  28
                                                                                          TeNeT Group
                                                                                           IIT Madras
IMT-A Requirements from ITU *
  * Test environments are described in Report ITU-R M.2135
                                       Average spectral efficiency
             Test environment (1)                Downlink                              Uplink
                                             (bit/s/Hz/sector)                   (bit/s/Hz/sector)
               Indoor (10 Kmph)                          3                             2.25
         Microcellular (30 Kmph)                        2.6                            1.80
          Base urban (120 Kmph)                         2.2                             1.4
          High speed (350 Kmph)                         1.1                             0.7

                                       Cell edge spectral efficiency
                    Test environment (1)         Downlink                              Uplink
                                             (bit/s/Hz/sector)                   (bit/s/Hz/sector)
                          Indoor                        1.0                            0.70
                       Microcellular                   0.75                            0.50
                        Base urban                     0.60                            0.30
                        High speed                     0.40                            0.15
  K.Giridhar/IITM                      Key Elements in Mobile Broadband Access
                                                                                                        30
                                                                                                       TeNeT Group
LTE & 802.16m submissions to IMT-A                                                                      IIT Madras

                                              3GPP vs 802.16m numbers
                                                 (FDD Downlink)
                                                        Indoor         Microcell            Base         High
                     Sector Spectral Efficiency                                             Urban       Speed
                         ITU Requirement                  3.0              2.6              2.2          1.1
                           3GPP LTE-A                 4.8 / 6.6            4.5              3.0          3.9
                          IEEE 802.16m                   5.90             3.13              2.44        3.12


                                        VoIP capacity (users/sector/MHz) for FDD
                                                                 3GPP             802.16m               ITU
                                    Test Environment
                                                                DL / UL           DL / UL           requirement
                                         Indoor                  140 /             136 /
                                                                                                        50
                                                                  131               156
                                      Microcellular             80 / 74          75 / 100               40
                                     Base coverage
                                         urban                  68 / 69           70 / 94               40
                                       High speed               91 / 94          93 / 102               30
   K.Giridhar/IITM                                Key Elements in Mobile Broadband Access
                                                                                                                     31
                                                                                                  TeNeT Group
                                                                                                   IIT Madras
Towards a common platform …..


                                                                                                 LTE, LTE-A
                 IEEE 802.16d/e/m
                                                                       Bandwidth         3        5      10       15      20
 Bandwidth           1.25   2.5   5     10     20                      (MHz)
  (MHz)

                                                                       Sampling Rate     4.608    7.68   15.136   23.04   30.72
Sampling rate        1.4    2.8   5.6   11.2   22.4
     (MHz)                                                             (MHz)




                     Common (reference) clock can be derived !

                      define proper sequence of interpolators/decimators
                      simple, poly-phase implementation
                      choice of mother filter is important



   K.Giridhar/IITM                             Key Elements in Mobile Broadband Access
                                                                                                                          32
                                                                                        TeNeT Group
                                                                                         IIT Madras
Outline
   What is Co-Channel Interference (CCI)?
           What is the consequence of CCI?
           Methods to mitigate CCI  overview


   IMT-A Standards Specs
           OFDM is the bedrock of all future wireless standards
                  What is OFDM (and its variants)?
                  Why OFDM?
           Overview of IEEE 802.16m and 3GPP LTE


   R&D case studies in CCI mitigation
           Conjugate Data Repetition
           Interference Cancelling Block Modulation
           CCI aware Receiver Processing


   Summary



         K.Giridhar/IITM                      Key Elements in Mobile Broadband Access
                                                                                                      33
                                                                                                 TeNeT Group
                                                                                                  IIT Madras
Summary
            Interference Management is vital for 4G Wireless to succeed

            Will LTE-A deliver required performance? How to assess performance?

            Link level performance analysis
                   Uncoded BER versus Coded BER


            System level performance analysis
                   Impact of H-ARQ
                   Impact of best-band selection
                   Impact of scheduling
                       Prop-fair versus Round-robin
                   Impact of unequally loaded cell/sectors
                   Impact of finite buffers and TCP flow control


            What metrics to use?
                   Link level metrics not adequate

                   System level “Spectral Efficiency” (average mutual information)
                        Average cell spectral efficiency (bits/sec/Hz/cell)
                        Cell-edge spectral efficiency (for bottom 5% of users)
                        Impact of pico-cells and femtos on these numbers is very significant

                      Avg. Throughput or Avg. Spectral Efficiency?


            Pragmatic Rate Adaptation is the Key – Talk #2 tomorrow!
  K.Giridhar/IITM                                      Key Elements in Mobile Broadband Access
                                                                                                               34
                                                                       TeNeT Group
                                                                        IIT Madras




            Key Elements of Mobile Broadband Systems




                      Discussion




                         Indore, Jan. 27, 2012


K.Giridhar/IITM              Key Elements in Mobile Broadband Access
                                                                                     35
                                                                                 TeNeT Group
                                                                                  IIT Madras
PHY Layer Research on CCI Mitigation

           Interference Awareness  key aspect of cognition

           Interference Aware Processing

                  Coding for Interference Channels
                     CDR

                     ICBM




                  Game-theory for Interference Channels

                  Interference aware
                     Channel Estimation

                     MIMO combining

                     Symbol detection

                     Iterative decoding




                  Robust statistics, MMSE & biased estimation, subspace
                   techniques, and game theory will be the building blocks
 K.Giridhar/IITM                       Key Elements in Mobile Broadband Access
                                                                                               36
                                                                 TeNeT Group
                                                                  IIT Madras




                   Interference Aware
          Channel Estimation & Symbol Detection




K.Giridhar/IITM
IITM / Oct. 2009       Key Elements in Mobile Broadband Access
                              Wireless Research @ IITM
                                                                               37
                                                                       TeNeT Group
                                                                        IIT Madras
Parametric Channel Estimation




                                                                FFT based CE




       Parametric CE                                                   mLS based CE




    K.Giridhar/IITM       Key Elements in Mobile Broadband Access
                                                                                      38
                                                                    TeNeT Group
Nulling vs MRC                                                       IIT Madras
(with 2 Rx antennas)




 1 x 2 OFDMA system
 No. of subcarriers=512
 Length of CP=32
 Channel taps=4
 No. of interferers=1




                             Performance comparison of Subspace based CE
                             with “Nulling” and “MRC” combining

     K.Giridhar/IITM      Key Elements in Mobile Broadband Access
                                                                                  39
                                                                                 TeNeT Group
                                                                                  IIT Madras
OFDMA downlink CE
-- Impact on Nulling Performance




                  Performance comparison of different CE methods with “nulling”


    K.Giridhar/IITM                    Key Elements in Mobile Broadband Access
                                                                                               40
                                                               TeNeT Group
                                                                IIT Madras




                   Robust LLRs




K.Giridhar/IITM
IITM / Oct. 2009     Key Elements in Mobile Broadband Access
                            Wireless Research @ IITM
                                                                             41
                                                                        TeNeT Group
                                                                         IIT Madras
Co-channel interference in Cellular




                                                         Overall noise seen by the
                                                         subcarriers no longer remains
                                                         Gaussian


                          The main parameters of Interference are
                          •Interference Power
                          •Number of affected sub-carriers




  K.Giridhar/IITM
  IITM / Oct. 2009       Key Elements in Mobile Broadband Access
                                Wireless Research @ IITM
                                                                                         42
                                                                                                          TeNeT Group
                                                                                                           IIT Madras
    Interference models



     Middleton                             Contaminated Gaussian (CG)
     Class-A model
                                                                                      x2                                     x2
                                                                                      212                                    2 nb
                                                                                                                                 2
                                             f n ( x)  (1  )         1        e                       1       e
                                                1                    21
                                                                        2                                  2 nb
                                                                                                               2

   Alpha-Stable                           Contaminated Cauchy (CC)
   noise
                                                                                             x2
                                                                         1               
                                                                                             21                 
                                           f n2 ( x)  (1   )                                    
                                                                                               2
                                                                                    e
                                                                      2 12                                (   x 2 )
    Mertz Model                           Contaminated Laplacian (CL)

  There are 3 parameters for each noise                                                  x2                                    2
                                                                                                                                   | x|
                                                                       1                 21                1                  nb
                                          f n3 ( x)  (1   )                                     
                                                                                           2                                    2
  pdf:                                                                           e                                   e
 1 : Noise variance
    2
                           : % of                                   2     2
                                                                                                           2   2
                            affected                                         1                                  nb

 nb : Interference power
  2                         subcarrier
                            s

       K.Giridhar/IITM
       IITM / Oct. 2009                   Key Elements in Mobile Broadband Access
                                                 Wireless Research @ IITM
                                                                                                                                            43
                                                                                                            TeNeT Group
                                                                                                             IIT Madras
  Optimal LLRs



Contaminated Gaussian LLR (CGLLR)                                           Contaminated Cauchy LLR (CCLLR)
                                  ( y 1)2                  ( y 1)2                                                                              
                                  k                        k                                             ( y 1)2
                                                                                                            k                                      
                        1  e       21
                                        2
                                                    e        2 nb 
                                                                  2
                                                                                                  1  e       21
                                                                                                                  2                                
                                                                                                                                                 
                        2                    2 nb
                                                                    
                                                                                                  2                        
                                                                                                                                        ( yk  1) ) 
                                                                                                                                                  2

 LCG ( yk )  log 
                  
                                                                     
                                                                            LCC ( yk )  log                                                        
                                   ( y 1)2                ( yk 1)2 
                                  k                                                                     ( y 1)2                                
                        1  e       21
                                        2
                                                    e 2 nb   2                           
                                                                                                   1  e
                                                                                                             k
                                                                                                                21
                                                                                                                  2                                
                  
                         2                    2 nb
                                                                                                                                                 
                  
                  
                                                                     
                                                                     
                                                                                             
                                                                                                  2                                      2 
                                                                                                                                        ( yk  1) )
                                                                                                                                                     
                                                                                                                                                    


Contaminated Laplacian LLR (CLLLR)
                                                                                                   Gaussian LLR (GLLR)
                                                                       
                                  ( y 1)2                             
                                  k                        2 | y 1| 
                                                                                                                  2 yk
                        1  e       21
                                        2
                                                e
                                                            
                                                             k
                                                                                                   LG ( yk ) 
                    
                    
                         2                  2 nb
                                                   
                                                                        
                                                                                                                 
                                                                       
 LCL ( yk )  log                                                      
                                  ( y 1)2                             
                                  k                                   
                    
                    
                         1  e       21
                                        2
                                                   e      2 | y 1| 
                                                              k        
                    
                    
                    
                         2                    2 nb
                                                                       
                                                                        
                                                                        
                                                                       

      K.Giridhar/IITM
      IITM / Oct. 2009                                   Key Elements in Mobile Broadband Access
                                                                Wireless Research @ IITM
                                                                                                                                             44
                                                                                                                         TeNeT Group
                                                                                                                          IIT Madras
         Robust Initial LLRs



                                          Applying Generalised Likelihood rule                                                     1        
          CCLLR                                                                                                                y  1) 2   
                                                                                                          LGLR ( yk )  log       k         
                                                                                                                                   1        
                Fix the values of               nb
                                                 2and                                                                         y  1) 2
                                                                                                                                  k
                                                                                                                                             
                                                                                                                                             

                                                                                                                Re-introducing  in the prev
Parametric Cauchy LLR (PCLLR)                                                                                   expression
                              ( y 1)2                           
                              k                                            Parametric Generalised Likelihood Ratio (PGLR)
                     0.7 e       21
                                    2         0.3*1.06 
                                       
                    2                 1.06  ( yk  1)2 ) 
                                                                  
LPC ( yk )  log 
                 
                                                                  
                                                                  
                                                                                                           p( yk | xk  1) 
                           
                              ( yk 1)2
                                                                                 LPGLR ( yk )  log                      
                     0.7 e 212  0.3*1.06                                                              p( yk | xk  1) 
                                                                                                                           
                 
                    2                 1.06  ( yk  1)2 ) 
                                                                  
                                                                 
                                                                                     p( yk | xk  i)               1       ,0           1       t
                                                                                                              ( yk  xk )2         ( yk  xk )2
                                                                                                                          t,        1       t
                                                                                                                               ( yk  xk )2


           K.Giridhar/IITM
           IITM / Oct. 2009                                           Key Elements in Mobile Broadband Access
                                                                             Wireless Research @ IITM
                                                                                                                                                 45
                                                                               TeNeT Group
                                                                                IIT Madras
LTE Example  Intf. Alignment + Robust LLR




    SIR = 0dB                                                                         MMSE-IC




                                            Frequency
                                                                                        LLR



                            SIR = 3 dB                               Time


                                                                                    Turbo decoder




                MMSE-IC cannot null out the pilots since they are not precoded

  K.Giridhar/IITM                    Key Elements in Mobile Broadband Access
                                                                                              46
                                                                                     TeNeT Group
                                                                                      IIT Madras
Coded BER




                          Robust LLR does not have an error floor

                                                                 Gaussian LLR




                                                                              Robust LLR(PGLR)
             Modulation    QPSK
             Rate          ½
             Block length 256
             Iterations    7
             Channel       Ped A



  K.Giridhar/IITM
  IITM / Oct. 2009                  Key Elements in Mobile Broadband Access
                                           Wireless Research @ IITM
                                                                                                   47
                                                                     TeNeT Group
                                                                      IIT Madras




                    Other Research Examples
                  from Wireless Group @ IITM




K.Giridhar/IITM            Key Elements in Mobile Broadband Access
                                                                                   48
                                                                   TeNeT Group
                                                                    IIT Madras




                   Pilotless Channel Tracking




K.Giridhar/IITM
IITM / Oct. 2009         Key Elements in Mobile Broadband Access
                                Wireless Research @ IITM
                                                                                 49
                                                                        TeNeT Group
                                                                         IIT Madras
 Pilotless Channel Tracking




Modified DDCT here
outperforms 12.5%
pilot-aided tracking




       K.Giridhar/IITM        Key Elements in Mobile Broadband Access
                                                                                      50
                                                                 TeNeT Group
                                                                  IIT Madras




                   Channel Shortening




K.Giridhar/IITM
IITM / Oct. 2009       Key Elements in Mobile Broadband Access
                              Wireless Research @ IITM
                                                                               51
                                                                            TeNeT Group
                                                                             IIT Madras



    Channel shortening for MIMO Systems

      Requirement of channel shortening:

             Over head of Cyclic prefix in OFDM system, reduces the
              bandwidth efficiency of system
             For a given wireless system the cyclic prefix is fixed, however the
              practical real channel is varying. The varying channel memory
              introduce IBI and ICI

      Solution:

              The elegant solution is to use in time domain pre-filter (called as
              channel shortening pre-filter at the receiver, which for reducing
              the channel memory




K.Giridhar/IITM
IITM / Oct. 2009                  Key Elements in Mobile Broadband Access
                                         Wireless Research @ IITM
                                                                                          52
                                                                           TeNeT Group
Nullspace Exploiting CSP                                                    IIT Madras


      MSSNR (Maximizing the shortened SNR)Based Approach
      MMSE (minimum mean Square)Based Approach




  L-Channel Order, N-CSP Order, Ncp- Cyclic Prefix
     K.Giridhar/IITM             Key Elements in Mobile Broadband Access
                                                                                         53
                                                                           TeNeT Group
                                                                            IIT Madras

 BER Comparision




Figure: Comparison of proposed approach with existing MMSE method 1x2 System
    Ncp=16, N=50, L=49 Nfft=512 BW=5MHz

    K.Giridhar/IITM              Key Elements in Mobile Broadband Access
                                                                                         54
                                                                TeNeT Group
                                                                 IIT Madras




                   2-phase Relays




K.Giridhar/IITM
IITM / Oct. 2009      Key Elements in Mobile Broadband Access
                             Wireless Research @ IITM
                                                                              55
      Node A      Relay R      Node B                            Node A                Relay R   TeNeTBGroup
                                                                                                    Node
          A                                                                                       IIT Madras
                    R            B                                    A                  R          B


      0                                                           0
      T                                                           T
     2T                                                        2T
    3T

    4T


     Figure 1: Data transfer in four phase relay (one-way)        Figure 2: Data transfer in two phase relay (Two-way)
     system                                                       system




K.Giridhar/IITM                              Key Elements in Mobile Broadband Access
                                                                                                                         56
                                                                                      TeNeT Group
                                                                                       IIT Madras




                  Figure -- Sum of average BER verses SNR using OPSK in a 2×2×2 MIMO AF relay model
                  for the U-P,RON, JSRP-M1,JSRP-M2,JSRP-M1-RA, and JSRP-M2-RA



K.Giridhar/IITM                           Key Elements in Mobile Broadband Access
                                                                                                      57
                                                                             TeNeT Group
                                                                              IIT Madras
Other Issues




           Phase synchronization among the nodes

           Extension to asynchronous channel gains

           Optimum constellation design

           Extension to multiuser and/or multihop channels

           Using other optimisation criteria

                   e.g., Minimum BER criterian



 K.Giridhar/IITM
 IITM / Oct. 2009                  Key Elements in Mobile Broadband Access
                                          Wireless Research @ IITM
                                                                                           58
                                                                                        TeNeT Group
                                                                                         IIT Madras




                  Figure -- Shows the comparison between minimum BER objective function and
                  SMSE objective function of various algorithms




K.Giridhar/IITM                           Key Elements in Mobile Broadband Access
                                                                                                      59
                                                                         TeNeT Group
                                                                          IIT Madras




                   Estimation with Mis-specified Statistics




K.Giridhar/IITM
IITM / Oct. 2009               Key Elements in Mobile Broadband Access
                                      Wireless Research @ IITM
                                                                                       60
                                                                             TeNeT Group
                                                                              IIT Madras



     Robustness of Linear Game-Theoretic Estimators

 Robust estimator compromises                                                     Opt Est
efficiency at optimal operating                                                     Rob Est
conditions as shown.
                                                                       MSE
 Since H estimator is the H
controller, its error dynamics is
stabler than that of the Kalman
filter.
 Stable error dynamics do not
guarantee lower MSE.
 Unstable filtering H operation
can be stabilized by increasing                                              Parameter

process noise covariance.


 K.Giridhar/IITM             Key Elements in Mobile Broadband Access
                                                                                              61
                                                                       TeNeT Group
                                                                        IIT Madras



    Results/Plots

1. The conditional dominance
property of the estimators
can be observed, depending
on the type of uncertainty.

2. By choosing , it is desirable
to operate at the least mSE
point.

3. Robust estimators can be
built as a function of  in real
operating conditions using
the properties of the
innovations sequence.

K.Giridhar/IITM              Key Elements in Mobile Broadband Access
                                                                                     62
                                                                     TeNeT Group
                                                                      IIT Madras




                   Tensor-based Subspace Estimation




K.Giridhar/IITM
IITM / Oct. 2009           Key Elements in Mobile Broadband Access
                                  Wireless Research @ IITM
                                                                                   63
                                                                    TeNeT Group
                                                                     IIT Madras



    Problem

   Multidimensional Parameter Estimation
   A single or a very few observations available
   Therefore Spatial Smoothing has to be applied
   Matrix based Parameter Estimation techniques provide good
    performance
   However, performance depends on spatial smoothing
    parameters
   Tensor-based Parameter Estimation techniques exhibit
    superior performance
   In addition, performance is almost independent of spatial
    smoothing parameters




K.Giridhar/IITM           Key Elements in Mobile Broadband Access
                                                                                  64
                                                            TeNeT Group
                                                             IIT Madras



    Result




K.Giridhar/IITM   Key Elements in Mobile Broadband Access
                                                                          65
                                                                          TeNeT Group
                                                                           IIT Madras




            Key Elements of Mobile Broadband Systems



                            Thank You!




                     CIT, Coimbatore , December 8, 2011


K.Giridhar/IITM                 Key Elements in Mobile Broadband Access
                                                                                        66

								
To top