reprt final

Upload: souhridsmohan

Post on 04-Jun-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/13/2019 Reprt Final

    1/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 1

    Chapter I

    INTRODUCTION

    1.1 Scope and Introduction

    The radio frequency spectrum is filled with noise and other signals, especially those

    frequencies where unlicensed transmitter operation under FCC part 15 rules is allowed. When using

    a wireless remote control system it is desirable to have a way of filtering out or ignoring those

    unwanted signals to prevent false data from being received. A simple way to accomplish this is to

    use an encoder IC at the transmitter and a decoder IC at the receiver. The encoder generates serial

    codes that are automatically sent three times and must be received at least twice before data is

    accepted as valid by the decoder circuit. In the early days of radio control", before these coding

    ICs were available, radio controlled garage doors sometimes opened themselves when they

    received transmissions from a plane passing overhead or a two-way radio operating in the

    area. Encoding and decoding is now used in most wireless control systems to prevent this

    type of interference.

    The complexity and the latency, which are often higher during the decoding process,

    make up the major decoding drawbacks in the current digital systems. The decoding data rate

    depends of the parallelism of the chosen architecture, i.e., its ability to process

    simultaneously multiple data. The architecture presents a high level of parallelism more it is

    able to achieve a very high data rate. In electronic design, the complexity of the circuit (i.e.,

    its surface) is often considered as a critical parameter. In this context of very high data rate,

    operating speed must be maximized with limitation of complexity induced by the parallelism

    of calculations. We will specially study the decoding of the codes belonging to the OSMLD

    (one step majority logic decoding) codes family and used in an iterative decoding process.The majority logic decoding uses a linear combination of a reduced set of syndromes

    represented by orthogonal equations. Hence it allows, by choosing a pipeline and / or

    parallelized architecture, to have a less complex decoding circuit with a high data rate.

  • 8/13/2019 Reprt Final

    2/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 2

    Chapter II

    LITERATURE SURVEY

    2.1 VHDL Design and FPGA Implementation of Weighted Majority Logic

    Decoders

    In this work, us propose a design and FPGA (Field Programmable Gate Arrays)

    implementation of two parallel architectures for majority logic decoder of low complexity for

    high data rate applications are proposed. These architectures are hard decision architecture

    (Hard in - Hard out (HIHO)) and the SIHO threshold decoding. The code used is the

    Difference Set Cyclic code (DSC (21, 11)). The VHDL (Very high speed integrated circuitHardware Description Language) design and the synthesis of such architecture shows such

    decoders can achieve high data rate with low complexity

    Fig 2.1: Parallel implementation of the proposed architecture

    LD for DSC (21, 11) code

  • 8/13/2019 Reprt Final

    3/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 3

    The complexity and the latency, which are often higher during the decoding process,

    make up the major decoding drawbacks in the current digital systems. The decoding data rate

    depends on the parallelism of the chosen architecture, i.e., its ability to process

    simultaneously multiple data. The architecture presents a high level of parallelism more it is

    able to achieve a very high data rate. In electronic design, the complexity of the circuit (i.e.,

    its surface) is often considered as a critical parameter. In this context of very high data rate,

    operating speed must be maximized with limitation of complexity induced by the parallelism

    of calculations. After reminding the principle of majority logic decoding, this paper will

    present the proposed the work with a hard decision and architecture of Massey's APP (A

    Posterior Probability) algorithm. Both architectures use the joint exploitation of parallelism of

    symbols and sub-blocks to achieve high data rate.

    The algorithmic structure of the threshold decoding is summarized as follows:

    For each j = n,,1

    Calculate the terms Bi and Wi, i in {1,..,J}

    Calculate Bo et Wo

    Calculate the extrinsic information W (yj)

    The output decoder is obtained by

    LLR(yj )=4 ES yj +W(yj)Decision

    {

    Summary:

    VHDL design and FPGA (Field Programmable Gate Arrays) implementation of twoparallel architectures for majority logic decoder is proposed.

    The architectures are hard decision architecture and the SIHO threshold decoding. Aim is to obtain low complexity for high data rate applications. The code used is the Difference Set Cyclic code. In electronic design, the complexity of the circuit is a critical parameter and operating

    speed must be maximized with limitation of complexity induced by the parallelism of

    calculations.

    The architecture of the threshold decoding has three fundamental units.

  • 8/13/2019 Reprt Final

    4/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 4

    The shift register contains the quantified symbols received in parallel, the loading ofthe register is controlled by input on the first rising front of the clock.

    In the implementation of the add-min operator and the adder, the number of inputports must be configurable as well as the number of bits per port to avoid poorperformance of the correction errors.

    The results of the add-min operation, the addition and the decision are available at theoutput at each rising edge of the clock.

    The approach of the VHDL design and FPGA implementation of the decoderproposed allow to achieve very high data rate with reducing complexity.

    A combinatorial architecture working in pipelined and parallelized mode has beenadopted.

    The hard decision decoder responds on the first rising edge of the clock with acomplexity of 173 LEs, for threshold decoding requiring 22 clock cycles so as that the

    code word is decoded with a complexity of 259 LEs.

    The decoder can be used on a turbo process in order to achieve a high data rate turbodecoder with a reasonable complexity.

    Fig 2.2: Parallel implantation of proposed architecture

    of the threshold decoding of MLC code

  • 8/13/2019 Reprt Final

    5/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 5

    2.2 An FPGA Implementation of Array LDPC Decoder

    Low-Density Parity-Check (LDPC) code is one kind of prominent error correcting codes

    (ECC) being considered in next generation industry standards. This paper presents an FPGA

    implementation of array code based Low-Density Parity- Check code decoder. The

    advantages of the proposed architecture as compared to the fully parallel or partially parallel

    architecture are: less memory requirement, avoidance of complex global interconnects and its

    satisfying throughput. These advantages are derived from exploiting the well-defined

    structure of the parity check matrix of array code based LDPC codes. Iterative decoding

    algorithms, also known as the message-passing algorithms are used for decoding LDPC

    codes.

    LDPC codes, due to their near-Shannon limit performance under iterative soft

    decoding have received significant attention. LDPC codes have been recently selected by the

    digital video broadcasting standard and are being seriously considered in various real-life

    applications such as magnetic storage, 10 GB Ethernet, and high-throughput wireless local

    area network. In the past few years, significant improvement in LDPC code construction and

    performance analysis has been experienced. In addition, several efforts are given to the

    implementation of LDPC code decoder. In general, the complex routing, large message

    memory requirement and limited throughput are the main obstacles in designing an LDPC

    decoder. This paper presents an efficient FGPA implementation for array based LDPC

    (ALDPC) codes, which can achieve a good trade-off between hardware complexity and

    decoding speed. Typically, over 70% memory can be reduced and 117Mbps throughput can

    be achieved. LDPC codes can typically be defined by an MxN parity check matrix H. The

    symbol N, represents the length of the block (i.e. the number of bits in the codeword), while

    the symbol M, represents the number of parity checks in the code. The rate of such a code is

    thus at least (N-M)/N. The LDPC code can be represented by a bipartite graph (Tanner graph)of variable nodes and check nodes as shown in Fig.

  • 8/13/2019 Reprt Final

    6/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 6

    Fig 2.3. Tanner graph for a regular (2, 3) LDPC code

    The decoder mainly consists of CPUs and VPUs. 47 (p) VPUs and 188 (M) CPUs are

    instantiated in the decoder. The CPUs are separated into four groups. CPU have four groups

    Each group has 47 members (CPU). Group 1 contains from CPU-1 to CPU-47, in charge of

    the row operation of row 1~47. Group 2 contains from CPU-48 to CPU-94, in charge of the

    row operation of row 48~94. Group 3 contains from CPU-95 to CPU-141, in charge of the

    row operation of row 95~ 141. Group 4 contains from CPU-142 to CPU-188, in charge of the

    row operation of row 142~188. There is no communication between groups. Except the

    CPUs in Group 1 (CPU-1iCPU-47), every 47 CPUs in a group form a chain, and each CPU

    communicates with its two neighbors.

    Channel

    Information

    Decoder bits

    Fig 2.4: Architecture of the decoder

    VPU

    CONTROL

    UNIT

    CPU GROUP 4

    CPU GROUP 3

    CPU GROUP 1

    CPU GROUP 2

  • 8/13/2019 Reprt Final

    7/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 7

    Summary:

    Low Density Parity Check (LDPC) code is an error correcting code. FPGA implementation for array based LDPC codes helps achieve a trade-off between

    hardware complexity and decoding speed.

    Over 70% memory can be reduced and 117Mbps throughput can be achieved. Calculations are independent between variable nodes or check nodes at each iteration. LDPC codes are suitable for parallel implementation. Low logic consumption

    2.3 Design and FPGA Implementation of Stochastic Turbo Decoder

    Turbo codes are a recent development in the field of forward-error-correction channel

    coding. The codes make use of three simple ideas: parallel concatenation of codes to allow

    simpler decoding; interleaving to provide better weight distribution; and soft decoding to

    enhance decoder decisions and maximize the gain from decoder interaction. Turbo codes

    immediately achieved performance results in the near range of the theoretically best values,

    giving a less than 1.2-fold overhead.

    Fig 2.5: The generic turbo encoder

  • 8/13/2019 Reprt Final

    8/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 8

    Stochastic decoding that is inspired by stochastic computation is an alternative

    technique for decoding of error-correcting codes. The extension of this approach to decode

    convolutional codes and turbo codes is discussed in this article. The switching activity

    sensitivity is circumvented and the latching problem is reduced by transforming the stochastic

    additions into stochastic multiplications in the exponential domain and using multiple streams

    with deterministic shufflers. The number of decoding cycles is thus considerably reduced

    with no performance degradation. Stochastic decoding, previously applied to the decoding of

    LDPC codes, can now be applied to decoding of turbo codes. In addition, the first hardware

    architecture for stochastic decoding of turbo codes is presented. The proposed architecture

    makes fully-parallel turbo decoding viable on FPGA devices. Results demonstrate the

    potential of stochastic decoding to implement fully-parallel turbo decoders.

    Principles of stochastic computation were elaborated in the 1960s by Gaines and

    Poppelbaum et al. As a method to carry out complex operations with a low hardware

    complexity. The main feature of this method is that the probabilities are converted into

    streams of stochastic bits using Bernoulli sequences in which the information is given by the

    statistics of the bit streams. Complex arithmetic operations on probabilities such as

    multiplication and division are transformed into operations on bits using elementary logic

    gates. Stochastic decoding of Forward Error Correction (FEC) is inspired by these principles.The two main appealing features of this approach for iterative decoding are very simple

    hardware structures of computing nodes and high-throughput decoding. Early stochastic

    methods were only successful for decoding special short/acyclic Hamming or LDPC codes .

    These methods result in poor decoding performance when used to decode practical codes on

    factor graphs (with cycles). An improved stochastic decoding approach was then proposed to

    decode practical LDPC codes. When compared with conventional Sum-Product

    implementations, stochastic decoding could provide near-optimal performance for practical

    LDPC codes.

    The potential of this method for low- complexity decoding was validated by an

    FPGA implementation of a (1024, 512) LDPC decoder. This first implementation provides a

    throughput of 706 Mbps at a Bit Error Rate (BER) of 106 with performance loss of about 0.1

    dB, compared to the floating-point BP, while occupying only 36% of a Virtex-4 LX200

    FPGA device . High data rates and architecture efficiencies were achieved thanks to many

    recent improvements. In addition, this approach was extended to well-known linear block

    codes with high-density parity-check matrices, namely BCH codes, Reed Solomon codes and

  • 8/13/2019 Reprt Final

    9/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 9

    product codes. Moreover, stochastic decoding is also of high interest for low power decoders

    and fault-tolerant Nano scale circuits. Turbo codes are a family of FECs that are especially

    attractive for mobile communication systems and have been adopted as part of several

    channel coding standards for high data rates such as UMTS and CDMA2000 (third-

    generation) or 3GPP-LTE (the last step toward the fourth generation). A major challenge in

    the implementation of turbo decoders is high-throughput decoding. Indeed, the next

    generations of mobile communication systems will require data rates of 1 Gb/s and beyond.

    Designers try to exploit the maximum feasible amount of parallelism in turbo decoders for

    the sake of higher throughput.

    However, due to the lack of parallelism in the MAP-based decoding algorithm, the

    implementation of fully-parallel turbo decoders is still challenging for practical turbo codes.

    Another form of stochastic algorithm was proposed in and used for trellis decoding algorithm

    of an acyclic (16, 11) Hamming code and a (256,121) product Turbo code based on 32

    component decoders of this Hamming code. An improved stochastic decoding approach was

    recently introduced to decode convolutional codes and turbo codes. The results provided in

    this paper validate the potential of stochastic decoding as a practical approach for high-

    throughput turbo decoders and encourage to keep on investigating in this way. One major

    problem in stochastic decoding which deeply degrades the performance is related to thesensitivity to the level of random switching activity. Different solutions such as Edge

    Memories (EMs) have been suggested to solve this latching problem. An original idea is

    introduced to reduce the number of clock cycles required to decode one code word and to

    replace the EMs by simple deterministic shufflers. The proposed architecture, that uses

    multiple streams in parallel to represent the same probability, presents the best architecture

    efficiency defined as the ratio between data throughput and hardware complexity.

    Summary:

    Stochastic decoding is an alternative technique for decoding of error-correcting codes. The switching activity sensitivity is circumvented and the latching problem is reduced

    by transforming the stochastic additions into stochastic multiplications in the

    exponential domain and using multiple streams with deterministic shufflers.

    The probabilities are converted into streams of stochastic bits using Bernoullisequences in which the information is given by the statistics of the bit streams.

  • 8/13/2019 Reprt Final

    10/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 10

    The two main features of this approach for iterative decoding are very simplehardware structures of computing nodes and high-throughput decoding.

    Turbo codes are a family of FECs that are especially attractive for mobilecommunication systems. Exploitation of the maximum feasible amount of parallelismin turbo decoders is preferred for the sake of higher throughput.

    One major problem in stochastic decoding which deeply degrades the performance isrelated to the sensitivity to the level of random switching activity.

    The proposed architecture uses multiple streams in parallel. The number of decoding cycles is considerably reduced with no performance

    degradation by transforming the stochastic additions into stochastic multiplications in

    the exponential domain and using multiple streams with deterministic shufflers. The proposed architecture makes fully-parallel turbo decoding viable on FPGA

    devices.

    2.4 VHDL Design and FPGA Implementation of a fully parallel BCH SISO

    Decoder

    In this work, a design and (FPGA) implementation of architecture of an entirely

    parallel SISO decoder for turbo decoding of the product codes with low complexity for highdata rate applications. As an example we study a soft input/output decision decoding for the

    BCH (31, 26, 3) code. The VHDL design and synthesis of such architecture showed that the

    use of the structure combining the sub-blocks parallelism with the symbols parallelism for

    the establishment of such decoder can achieve a data rate of 75.5 Gb/s with low complexity,

    about 2199 CLBs.

    The turbo codes in blocks (TCB) are an alternative to convolutional turbo codes

    introduced in 1993 by C. Berrou, (TCBs) have been proposed by R. Pyndiah in 1994, the

    decoding uses weighted (Soft) input/output SISO algorithm. The current digital systems, the

    channel coding especially decoding, require high data rate rates. The decoder data rate

    depends on the parallelism of the chosen architecture, i.e. its ability to process multiple data

    simultaneously. More the architecture provides a high level of parallelism, more it is able to

    achieve very high speed. In electronic design, the complexity of the circuit (i.e, its surface) is

    often considered as a critical parameter. In this context of very high speed operating speed

    must be maximized, with limitation of complexity induced by the parallelism of calculations.

  • 8/13/2019 Reprt Final

    11/20

  • 8/13/2019 Reprt Final

    12/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 12

    The decoder works in soft decision way. The entry consists of analog samples (called

    symbols) encoded on q=5 bits: 4 bits for reliability and 1 for sign. The sign bit represents the

    binary value of the symbol ('0 'or '1'), the absolute value of a received bit represent its

    reliability (high reliability means that this bit has more chance to be correct). The implemented

    decoder uses the Chase-Pyndiah algorithm.

    The Chase-Pyndiah algorithm consists of three parts: The first section of the

    algorithm is extracting some useful parameter from the received word (reception part).

    Access to data is performed in parallel; the implementation of this part is done in parallel, this

    presupposes that the last part of the algorithm (transmission part) is also implemented in

    parallel. The intermediate part, named processing part occurs at the level of test vectors. This

    part can be implemented according to different degrees of parallelism. The proposed

    architecture introduces the combinatorial approach to the implementation of each basic block

    (part). To implement this elementary decoder, we have to translated the steps of decoding (for

    the half iteration) in simple blocks.The sub-block of the receiving part works in parallel on

    the input symbols simultaneously. These operations are the computation of the syndrome and

    the sorting to get the least reliable components. We obtain all data on the first rising edge of

    the clock. One period allows to process all test vectors: calculating syndrome, decoding and

    computing metric for each hard decision related to test vector. Thereafter, since the decided

    vector and its competitor (s) are known, computing reliability can be done in one clock

    period, according to this the chosen architecture requires only 2-clock-cycle latency.

    Summary:

    A design and (FPGA) implementation of architecture of a fully parallel SISO decoderfor turbo decoding of the product codes with low complexity for high data rate

    applications is proposed.

    In this, joint exploitation of parallelism symbol and parallelism of sub blocks are doneto achieve high data rate.

    Chase-Pyndiah algorithm is used in the proposed system. The decoder can reach some data rate superior to the Gbit/s with a reasonable

    complexity by limiting the degradation

    The sub-block of the receiving part works in parallel on the input symbolssimultaneously.

  • 8/13/2019 Reprt Final

    13/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 13

    These operations are the computation of the syndrome and the sorting to get the leastreliable components.

    All data on the first rising edge of the clock is obtained.

    One period allows to process all test vectors: calculating syndrome, decoding andcomputing metric for each hard decision related to test vector.

    The chosen architecture requires only 2-clock-cycle latency.2.5 A New Iterative Threshold Decoding Algorithm for One Step Majority

    Logic Decodable Block Codes

    The performance of iterative decoding algorithm for one-step majority logic

    decodable (OSMLD) codes is investigated. A new soft-in soft-out of APP threshold

    algorithm which is able to decode theses codes nearly as well as belief propagation (BP)

    algorithm. However the computation time of the proposed algorithm is very low. The

    developed algorithm can also be applied to product codes and parallel concatenated codes

    based on block codes. Numerical results on both AWGN and Rayleigh channels are provided.

    The performance of iterative decoding of parallel concatenated code (17633,8595) with rate

    0.5 is only 1.8 dB away from the Shannon capacity limit at a BER of 10-5.

    TURBO codes or LDPC codes are currently the most effective solution for Forward

    Error-Correction (FEC) in applications requiring either high code rates (R > 0.8), very low

    error floors, or low-complexity decoders able to operate at several hundreds of megabits per

    second and even higher. Practical implications of theses codes are numerous namely 3G

    wireless phones, Digital Video Broadcasting (DVB) systems, or Wireless Metropolitan Area

    Networks (WMAN). Turbo decoding relies on the exchange of probabilistic messages (the

    so-called extrinsic information) between two soft-input soft-output (SISO) decoders for the

    component convolutional codes. However, for concatenated schemes with block component

    codes, the computational complexity of trellis-based SISO decoding algorithms is often high.

    This has led to look for new SISO decoding algorithms with low complexity and high

    performance. These algorithms calculate extrinsic information using classical decoders such

    as Chase algorithm, ordered statistics decoding algorithm and Hartmann/Rudolph algorithm.

    In this perspective we present a new iterative decoding algorithm based on a SISO extension

    form of threshold algorithm.

  • 8/13/2019 Reprt Final

    14/20

  • 8/13/2019 Reprt Final

    15/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 15

    A new iterative threshold decoding algorithm for simple codes, product codes andparallel concatenated block codes based on one-step majority logic decodable code.

    We use an extension of Masseys algorithm as a Soft input/ Soft-output componentdecoder. The structure of our iterative decoder follows the model of Pyndiah withsome modifications.

    This algorithm has been tested on several codes based on OSMLD codes and goodperformances have been obtained over the Gaussian and Rayleigh fading channels. It

    is interesting to extend this iterative decoding algorithm on quasi- cyclic and multi-

    step majority logic decodable codes.

  • 8/13/2019 Reprt Final

    16/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 16

    CHAPTER III

    WORKTO BE DONE

    3.1 Introduction

    Over the years of experimentation and research in regards to FPGA implementation of

    weighted majority decoders; The significance of the combinational architecture working in

    pipeline and parallelized mode has resulted in very high data rate and reduced complexity.

    Since complexity and high data rate are critical parameters in electronic design, even the

    above proposed method was an ideal one.

    Therefore our project comprises of a SISO architecture model in addition to the other two

    architecture models. So that the working model can be much more efficient in regards to

    speed optimization and complexity.

    3.2 Proposed work

    3.2.1 Design of SISO threshold decoding

    The basic idea is to modify the conventional threshold algorithm by associating a reliability

    to each decision exploited in iterative decoding

    Y_rec Extrinsic information

    Decision function

    Fig 3.1: Implementation of SOFT output decoder

    3.2.2 Implementation of the parallel SISO threshold decoder of DSC code

    Several studies on the "SISO decoders" have already been carried out. In this paper theVHDL description of the SISO threshold decoding is made so that each unit of the proposed

    architecture is described in an independent entity. The final architecture of the parallel SISO

    threshold decoding is given in figure 3.2. The proposed architecture has been described in

    VHDL and embedded on FPGA using the Alteras Quartus II tool. The FPGA circuit used is

    similar to the one used toold decoder, the operations are obtained on the first rising edge of

    the clock and the occupied area for this decoder is 3672 LEs (logic elements) with 190

    inputs/outputs for the DSC (21, 11) code.

    SISO threshold

    decoder

  • 8/13/2019 Reprt Final

    17/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 17

    Fig3.2 : Parallel Implementation of proposed architecture of the SISO

    3.2.3 Iterative Threshold Decoding

    A. Iterative threshold decoding

    The iterative decoding of codes OSMLD was introduced for the first time by Lucas. Two

    years after, the latter with Fossorier et al. have proposed the belief propagation algorithm for

    decoding of OSMLD codes, They showed that the iterative Hartmann-Rudolf-Lucas

    decoding, originally used for the decoding of product codes can be used to decode the codes

    OSMLD. In the process of iterative decoding, each decoder takes advantage of the extrinsic

    information produced by the other decoder in the previous step. How extrinsic information is

    transmitted and the way it is exploited by decoders to make their decision are not yet closed.

    The iterative decoding process that we use follows the Pyndiahs scheme.

  • 8/13/2019 Reprt Final

    18/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 18

    Fig 3.4:Component decoder

    B. Implementation of parallel iterative threshold decoding

    The final design of the parallel iterative decoder is de- scribed in VHDL and embedded on

    FPGA using the Alteras Quartus II tool. For the complexity of the global entity, the area

    occupied by the fully parallel iterative show in table 3.1 for the two codes DSC (21, 11) and

    DSC (73, 45) with 4 and 5 bits quantification. We were able to simulate and verify the

    decoding of 2000 words used for each SNR following a simulation pro-tocol shows the

    performances of our iterative decoder using the quantified data of 5 bits and (q) fixed for the

    DSC(21, 11) (resp. DSC(73, 45)) code.

    Table3.1

  • 8/13/2019 Reprt Final

    19/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 19

    Summary:

    The complexity and the latency are often higher during the decoding process, creatingmajor decoding drawbacks in the current digital systems.

    In electronic design, the complexity of the circuit is often considered as a criticalparameter.

    Operating speed must be maximized with limitation of complexity induced by theparallelism of calculations for high data rate.

    The architecture presents a high level of parallelism more than it is able to achieve avery high data rate.

    The architectures described can be used in digital systems for high data rate andreduced complexity.

    Can be incorporated with digital systems to increase efficiency of data decoding.

  • 8/13/2019 Reprt Final

    20/20

    VHDL DESIGN AND FPGA IMPLEMENTATION OF A FULLY PARALELL ARCHITECTURE FOR ITERATIVE DECODER

    OF MAJORITY LOGIC CODES FOR HIGH DATA RATE APPLICATIONS

    Dept Of ECE,KMCTCE Page 20

    REFERENCE

    [1] Hagenauer, E. Offer, and L. Papke, "iterative decoding of binary block and

    convolutional codes", IEEE Trans. Inform. Theory, Mar. 1996, Vol. 42, pp. 429-446

    [2] R. Lucas, M. Bossert and M. Breitbach, "On Iterative Soft-Decision Decoding of

    Linear Binary Block Codes and Product Codes", IEEE Journal on selected areas in

    commu- nications, February 1998, Vol. 16, N 2, pp. 276-296.

    [3] R. Pyndiah, Near optimum decoding of product codes: Block Turbo Codes", IEEE

    Trans. on Comm., vol 46, n8, August 1998, pp. 1003-1010.

    [4] R. Lucas, M. P. C. Fossorier, Yu Kou, and Shu Lin, "Iterative Decoding of One-StepMajority Logic Decodable Codes Based on Belief Propagation", IEEE Trans. Commun,

    Vol. 48, No. 6, pp.931-937, June 2000.

    [5] M. Belkasmi, M. Lahmer, F. Ayoub Iterative Threshold Decoding of Product Codes

    Constructed From Majority Logic Decodable Codes , 2nd International Conf. on Infor-

    mation and Communication Technologies : From Theory to Applications, IEEE

    ICTTA06, Apr. 2006, Damascus, Syria, pp. 2376 2381.