MIMO with MUMS

Document Sample
MIMO with MUMS Powered By Docstoc
					MIMO with MUMS

2E1367 Project Course in DSP

                           Group Red 04
                              2004.5.28
Outline
 Introduction                          Implementation
     MIMO systems
     MUMS
 Theory for 3 MIMO
  schemes                               Results
     Weighted Orthogonal Space-time
      Block Coding
     Differential Space-time
      Modulation
     Spatial Multiplexing
 System Model
Introduction
 MIMO systems
     Multiple Input Multiple Output communication systems for
      a higher data rate and better quality.
 MUMS
     A test-bed with four radio nodes (Tx1,Tx2, Rx1,Rx2)
     Two antennas each node
     WIDELAB radio equipment
Theory - WOSTBC
 Weighted Orthogonal Space-time Block Coding
  (WOSTBC)
 Alamouti Space-time Coding-----encoder
         h11

  Tx1 h21           Rx1
                                      TX1    TX2
                           Time t     S0     S1
        h12
                          Time t +T   -S1*   S0*
              h22
  Tx2               Rx2
Theory - WOSTBC
 Alamouti Space-time Coding----Receiver

                                        RX1        RX2
        h11
                           Time t        r0         r2
 Tx1 h21           Rx1   Time t +T       r1         r3
       h12
                         r 0  h11 s0  h12  s1  n0
             h22
 Tx2               Rx2   r1  h11 s1 *  h12  s0 *  n1
                         r 2  h21 s0  h22  s1  n2
                         r 3  h21 s1 *  h22  s0 *  n3
Theory - WOSTBC
 Alamouti Space-time Coding----Decoder

  s0  h11* r 0  h12  r1*  h21* r 2  h22  r 3 *
  ˆ
  s1  h12 * r 0  h11 r1*  h22 * r 2  h21 r 3 *
  ˆ
  s0  (| h11 |2  | h21 |2  | h12 |2  | h22 |2 )  s0  N 0
  ˆ
  ˆ1  (| h11 |2  | h21 |2  | h12 |2  | h22 |2 )  s1  N1
  s
       r 0 r1
                       Decoder                s 0, s1
                                                 ˆ ˆ
       r 2 r 3
Theory - Weighting
 Goal: To maximize the SNR at the receiver
 Idea: Put more power on the transmitter that
  has a higher gain of channels
   If                                             h11

   h11  h21        h12        h22
        2      2           2           2
                                           Tx1 h21           Rx1

            Allot full power to TX1              h12

                                                       h22
                                           Tx2               Rx2
  VICE VERSA
   Theory –Differential Space-time Modulation
   Modulation                 Optimal Unitary Group Codes (n = t= 2 )

    Optimal Unitary Group
    Codes                          R             M                G

                                                              1 0 
                                   0.5         BPSK           0  1
                                                                   

                                                            j 0  0  1
Bits block                 G       1.5         QPSK        0  j , 1 0 
                                                                       
               Modulator
                                                           w8 0  0  1
                                   2            8PSK              ,    
                                                            0 w*8  1 0 
                                                                  




                 j / 4
   Notation:w8  e
Theory –Differential Space-time Modulation
 Differential Encoding
                                 1  1
  X0  D                       D
                                 1 1 
  X k  X k 1Gk
                          Gk                 Xk


                                      Z 1
Theory –Differential Space-time Modulation
 Differential Decoder


           GG
                       
     ˆ  arg max Re Tr GY Y
     G                   k  k 1   
                                       Re TrG1




                                                          Choose the G matrix
                                                          Maximize the value
                                   .                  .
                       
Yk                   Yk Yk 1      .                  .                         ˆ
                                                                                G
           {} +                    .                  .
                                   .                  .
                                   .                  .

           Z 1                        Re TrG m 
Theory – Spatial Multiplexing
 Utilize the feedback information of channel condition
 Water Filling: maximize the information capacity

    Receiver:                                             H
                                     S1                                    ˆ
                                                                           s1
    r  HWs  v  U H  H W s  v            W                        G
                                     S2                                    ˆ
                                                                           s2
                                                     TX       RX
            s  Gr
                                          Singular Value Decomposition (SVD):
       s  H W s  U H v
                        *
                                                                   
                                                  H  U H  H VH
                                                                       
                                                  W  UW W VW
Theory – Spatial Multiplexing
 Solution of W
                                                    Start M=2

  U W  VH         VW  I
                                                    1                1
                                                    (1   2 i 1 2 )
                                                                M

                                                                     H ,i
        W ,max 0                                 M
   W  
         0      W ,min 
                            M  M 1                    2
                                             
                                             2
                                                       2 , i  1,2
                                             W ,i
                                                         H ,i
 Solution of G
                                        No             W ,M  0
    G U
                                                        2
               *
               H
                                                               Yes
                                                         End
Theory – Spatial Multiplexing
 Adaptive modulation over time and space
 QoS Based Water Filling
    QoS: BER threshold 
    Mapping function of SNR to BER:                 i  F ( i , M i )


                                                      W ,i H ,i
                                                        2     2

    Step 1:       W ,i  WF ( H )                i 
                                                         2

    Step 2:        i  F ( i , M i )          Mi



Notation: M i: modulation order               i : SNR                i : BER
System Model
Transmitter

     Modulation       Frame            Up            Pulse              Up
                     Packing         sampling       shaping          conversion




Receiver

          Down           Matched        Sync      down         Frequency
        conversion        filter                sampling      offset comp




                               Demodulation      Detection       Channel
                                                                Estimation
Implementation - Frame packing (TX)
 Three kinds of frames                                BPSK
    Sync frame                                   Sync + freq_offset

                             32 tr symb
                                         BPSK alamouti
    Training frame                   Modulation identification

           16 tr symb        2 mod_id symb            14 zero symb
        BPSK orthogonal                              Noise estimation
        Channel estimation

    Data frame                  BPSK / QPSK / 16QAM

                             32 data symb
Implementation - Frame packing (RX)
 Feedback buffer
    First byte                  Unused


        7      6   5    4   3      2    1      0


  MOD_ID bits      WEIGHT_ID bits
  00    BPSK           00       TX1 more power
  01    QPSK           01       TX2 more power
  10   16QAM           10        Equal power
  11    Unused         11          Unused
Implementation - Program flow
                                       32 * 5
Feedback delay: 5 frames T frame         3.3ms
                                   48000
Feedback buffer size: 4 bytes
                                            Read 1st FB                                     Read 2nd FB
     Send tr frame                         Send tr frame                                    Send tr frame
      sync                      Receive 1st FB                           Receive 2nd FB
TX      0    1   2      3   4     5   6     7      8   9   10      11   12   13   14   15   16   17   …




                     FB1                                        FB2



RX      0    1   2      3   4     5   6     7      8   9   10      11   12   13   14   15   16   17   …

     Estimate channel                           Estimate channel                        Estimate channel
         Write FB                                   Write FB                                Write FB
Implementation - Frequency offset I
 Least squares line fitting
                                                                                             Slope
                       j *( 2 *f *n   0 )   (n)
 r0 (n)  s0 (n) * e                             -0.4
                                                                the phase of training unwrapped




                                                -0.45


                             (n)
                                                 -0.5


                  Im[ r0 (n)]
 (n)  arctan(               )                 -0.55

                  Re[ r0 (n)]
                                                 -0.6



 f  0.002                                    -0.65
                                                        0   5   10       15        20         25   30   35


 (96 Hz )                                                                    n
Implementation - Frequency offset II
 Frequency offset compensation

                     Before                                                         After

                         Signal Space of 16QAM
                                                                                  Signal Space of 16QAM
    5                                                            4

    4
                                                                 3
    3
                                                                 2
    2

    1                                                            1


    0                                                            0

   -1
                                                                 -1
   -2
                                                                 -2
   -3
                                                                 -3
   -4

   -5                                                            -4
     -5   -4   -3   -2      -1     0     1       2   3   4   5     -4   -3   -2    -1       0       1     2   3   4
Implementation - Design parameters I
 SNR threshold

                BER
                        BPSK
                               QPSK
                                      16-QAM


     Target BER =10-3




                                               SNR



                          20dB         50dB
Implementation - Design parameters II
 Power weight threshold
                                                  Tx1 power            Tx2 power
          h11
  Tx1           Rx1   tot _ power _ received | h11 |2  | h21 |2  | h12 |2  | h22 |2

    h21
                                 Receiver                                  Transmitter
                         Tx1 _ power _ received
    h12                                          0.8                  90% power to Tx1
                         tot _ power _ received
  Tx2     h22   Rx2      Tx1 _ power _ received
                                                 0.2                  10% power to Tx1
                         tot _ power _ received

                                    else                               50% power to Tx1
Implementation - Real time issues
 Code optimization (processing speed and memory space)
    Complex number (LDDW)
    Floating point division (extract the exponent)
    Trigonometric functions (C67x FastRTS Library, Cisoid Table)
Results
 Throughput
                                     48000
  theoretical _max_ throughput            * no _ bits _ per _ symb
                                       5

                                   48000
  pratical _ max_ throughput            * no _ bits _ per _ symb *
                                     5
                                               67  1  9
  no _ bits _ per _ symb  1,2,4                         0.8507
                                                  67

                               no _ correct _ bits _ received
  measured _ throughput 
                                         tot _ time

                  67 * 32 * 5
   tot _ time                 223.3ms
                   48000
Results
 Throughput (cont’d)
  theoretical_max_throughput           BPSK: 9.6K bps (8.16K bps)
  (pratical_max_throughput)
                                       QPSK: 19.2K bps (16.3K bps)

                                       16QAM: 38.4K bps (32.6K bps)


  Measured_throughput (alamouti)              BPSK: 8.15K bps
                                              QPSK: 19.15K bps
                                              16QAM: 30.72K bps
  Measured throughput (alamouti with          32.02K bps
  adaptive modulation)
   Results
                                                                                                                SISO
                                                                                                                                                          MIMO
 Simulations
                                                                                                       -3
             -1       BER performance of adaptive modulation                                       x 10                          average data rate
            10                                                 WOSTBC
                                                               SISO
                                                               DSTM
                                                                                              10
                                                               Spatial Multiplexing
                                                                                                            adaptive MIMO
                                                                                              9             BPSK
                                                                                                            QPSK
             -2                                                                                             16QAM
            10                                                                                8
                                                                                                            adaptive SISO

                                                                                              7
      BER




                                                                                       Rate
                                                                                              6

             -3
            10                                                                                5

                                                                                              4

                                                                                              3

             -4
            10                                                                                2
                  5              10                     15                        20               5                        10                       15   20
                                          SNR(dB)                                                                                    SNR(dB)




             MIMO: lower BER                                                                           MIMO: higher data rate

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:9/1/2012
language:English
pages:25