Dirty paper code design using nested trellis

download Dirty paper code design using nested trellis

of 24

Transcript of Dirty paper code design using nested trellis

  • 7/30/2019 Dirty paper code design using nested trellis

    1/24

    Dirty Paper Code design based on

    nested lattice codeand

    Performance analysisBy:

    Krishna Prasad Phelu

    (067/MSI/609)

    (102139)

    1

    Date: 13 November, 2011.

  • 7/30/2019 Dirty paper code design using nested trellis

    2/24

    Presentation outlines

    Background

    Problem definition

    2

  • 7/30/2019 Dirty paper code design using nested trellis

    3/24

    Motivation

    Figure: channel coding with side information.

    Multi-terminal communication networks is one of

    the fields of interest nowadays.

    Channel coding with the side information (CCSI)

    3

  • 7/30/2019 Dirty paper code design using nested trellis

    4/24

    Motivation: CCSI

    1

    log( )2

    X

    S Z

    PC

    P P

    Capacity with side information (S) not known to both

    encoder and decoder.

    Capacity with S known to both encoder and decoder.

    1

    log( )2

    X

    Z

    PC

    P

    Costa[1] proved in 1983 with S known only to encoder

    1log( )

    2

    X

    Z

    PC

    P

    4

  • 7/30/2019 Dirty paper code design using nested trellis

    5/24

    Objective

    To research on the DPC (dirty paper code), a special

    case of CCSI.

    To design DPC based on nested lattice code using

    TCM/TCQ scheme.

    5

  • 7/30/2019 Dirty paper code design using nested trellis

    6/24

    Problem definition

    Power constraint on X:2

    1

    1 n

    i X

    i

    X Pn

    Received signal, Y=X+S+Z.

    S is not known at the decoder but known non-casually

    at encoder.

    (0, )Z

    Z N P Noise:

    Figure: DPC channel.

    6

  • 7/30/2019 Dirty paper code design using nested trellis

    7/24

    Problem definitionCont

    Tx can not simply pre-subtract S from X.

    Tx: X =X-S.Rx: Y = X + S + Z = X + Z

    Direct pre-subtraction have sever power penalty.

    power of X: E[X2]=PX+PS

    PS can be much higher than PX

    7

  • 7/30/2019 Dirty paper code design using nested trellis

    8/24

    Tomlinson-Harashima Precoding

    (THP) 1D implementation of Dirty paper coding problem

    Mod- operation is performed on X=(X-S)

    Tx: X mod =(X-S) mod .

    X is limited to interval

    [0, )X If THP has no power penalty.

    Equivalent to scalar quantization

    8

  • 7/30/2019 Dirty paper code design using nested trellis

    9/24

    Figure: Power penalty due to direct pre-subtraction for same signal.

    9

  • 7/30/2019 Dirty paper code design using nested trellis

    10/24

    Dirty paper coding (DPC)

    Consider an entire sequence of S for higher

    dimensional mod operation

    Joint source and channel code design problem.

    Zamir et al [4] proposed a practical nested code

    scheme for joint source and channel code design.

    TCQ/TCM is used.

    10

  • 7/30/2019 Dirty paper code design using nested trellis

    11/24

    DPC using TCQ/TCM

    X

    X Z

    P

    P P

    Figure: DPC Encoder using TCQ/TCM scheme.

    11

  • 7/30/2019 Dirty paper code design using nested trellis

    12/24

    DPC using TCQ/TCM Contd

    12

    C2 is the channel code (TCM)

    C1 concatenated with C2 serves as a source code

    (TCQ)

    TCQ uses Viterbi algorithm to quantize S to U

  • 7/30/2019 Dirty paper code design using nested trellis

    13/24

    DPC using TCQ/TCM Contd

    H is syndrome former of code C1. We use the viterbi decoder for decoding code from

    C2

    Figure: DPC Decoder using TCQ/TCM.

    13

  • 7/30/2019 Dirty paper code design using nested trellis

    14/24

    DPC using TCQ/TCM Contd

    Figure: A TCQ/TCM encoder (rate 1/2/3 encoder).

    14

  • 7/30/2019 Dirty paper code design using nested trellis

    15/24

    DPC using TCQ/TCM Contd

    The trellis structure change

    depending on the data bits.

    difference between the Viterbi

    algorithm and nested code

    15

    6

  • 7/30/2019 Dirty paper code design using nested trellis

    16/24

    DPC using TCQ/TCM Contd

    SF and ISF of C1 can be designed and implemented

    using following sequential circuit.

    Figure: Syndrome former of code C1 Figure: ISF of code C1

    16

    [ ,1]T TH D 1( ) [0,1]TH

  • 7/30/2019 Dirty paper code design using nested trellis

    17/24

    Figure: DPC system simulation using TCQ/TCM scheme.

    17

    8

  • 7/30/2019 Dirty paper code design using nested trellis

    18/24

    Figure: DPC performance.

    18

  • 7/30/2019 Dirty paper code design using nested trellis

    19/24

    Figure: Capacity gain that can be achieved with DPC.

    19

    20

  • 7/30/2019 Dirty paper code design using nested trellis

    20/24

    Recommendation

    For better performance of DPC

    Use convolutional codes with very high constraint

    lengths

    TCQ/TTCM scheme

    Turbo-like TCQ/TTCM scheme

    20

    21

  • 7/30/2019 Dirty paper code design using nested trellis

    21/24

    Conclussion

    DPC is efficient interference pre-cancelation scheme at

    transmitter.

    DPC can be used to achieve channel capacity limit,

    without power penalty in Multi-terminal

    communication networks (like MIMO).

    DPC can be designed using nested code scheme.

    21

    22

  • 7/30/2019 Dirty paper code design using nested trellis

    22/24

    References:[1] M. Costa, Writing on dirty paper, IEEE Transactions on Information

    Theory, vol. It-29, pp. 439-441, May 1983.

    [2] Y. Sun, Y. Yang, A. D. Liveris, V. Stankovic, and Z. Xiong, "Near-Capacity Dirty- Paper Code Design: A Source-Channel CodingApproach", IEEE Transactions on Information Theory, vol. 55, no. 7,

    July 2009.[3] Yong Sun, "Source-channel coding for robust image transmission and

    for dirty paper coding", Ph.D. dissertation, Office of Graduate Studies ofTexas A&M University, Dec 2005.

    [4] R. Zamir, S. Shamai, and U. Erez, Nested linear/lattice codes for

    structured multiterminalbinning, IEEE Trans. Inform. Theory, vol. 48,pp. 1250-1276, June 2002.

    [5] J. Chou, S. Pradhan, and K. Ramchandran, Turbo coded trellis-basedconstructions for data embedding: channel coding with sideinformation, in Proc. 35thAsilomar Conf. Signals, Systems andComputers, Pacific Grove, CA, November 2001, vol. 1, pp. 305309.

    22

    23

  • 7/30/2019 Dirty paper code design using nested trellis

    23/24

    References:[6] Y. Sun, M. Uppal, A. Liveris, S. Cheng, V. Stankovic, and Z. Xiong,

    Nested turbo codes for the Costa problem, IEEE Trans.Communications, Vol. 56, pp. 1- 12,Janaury 2008.

    [7] M. A. Uppal, " Code design for multiple-input multiple-output broadcastchannel, MSc Thesis, Texas A&M University, August 2006.

    [8] P. Bhagawat, W. Wang, M. Uppal, G. Choi, Z. Xiong, M. Yeary and A.Harris, An FPGA Implementation of Dirty Paper Precoder, ICC 2007proceeding.

    [9] W. Yu, D. Varodayan, and J. Cioffi, Trellis and convolutional precodingfor transmitter-based interference pre-subtraction, IEEE Trans.Commun., vol. 53, no. 7, pp. 12201230, Jul. 2005

    [10] Zhenyu Tu, Jing Li, Blum, R.S., Compression of a Binary Sourcewith Side Information Using Parallelly Concatenated ConvolutionalCodes, vol. 1 pp. 46-50, January 2005.

    [11] www.wikipedia.com/ dirty paper coding.

    23

    24

  • 7/30/2019 Dirty paper code design using nested trellis

    24/24

    24