Coding And Decoding Algorithm For LDPC Codes And Its Implementation

Low Density Parity Check(LDPC)codes are one of the best error correction codes.The development of the Very Large Scale Integration(VLSI)reduces the implementation complexity of LDPC encoder and decoder.Thus LDPC codes have been widely used in various kinds of modern communication systems and became a hot research topic.This dissertation mainly focuses on the research of the decoding method,the decoder design and the construction method of the parity check matrix for LDPC codes.The major contents and achievements are concluded as follows:1.Decoding method.A simplified calculation method for decoding intrinsic messages was proposed.When the codewords were M-ary modulated and passed over an Additive White Gaussian Noise(AWGN)channel or a Rayleigh channel,the formulas for the intrinsic Log-Likelihood Ratio(LLR)messages were given respectively and the simplified approximate method was provided also.Simulation results show that the effects of this method on the decoding performance are negligible.A modified Min Sum decoding algorithm based on Linear Minimum Mean Square Error(LMMSE)estimate criterion was proposed and the fast algorithm of obtaining the corresponding estimated parameters was given.Comprehensive analysis and comparison demonstrate that the proposed algorithm has advantages including low computation complexity,simple calculation for estimated parameters and good decoding performance which is very close to BP algorithm and it can obtain a gain of 0.3-0.6 dB over other kinds of modified Min Sum algorithms.2.Decoder design.A non-uniform quantization method for LDPC decoding algorithm was put forward.The decoding performance is improved by ensuring the quantization precision of the low level data during the decoding,and the quantization length required is reduced by narrowing the quantization range and reducing the quantization precision of the high level data.So the method can save 2-3bit compared to using uniform quantization under the same level of BER performance.It not only save the storage resource but also reduce the complexity of the data exchange network and the arithmetic processing units.Furthermore a minimum-latency pipelined layered LDPC decoder was designed and implemented.A solution to the problem of the data updating conflict in pipelined layered LDPC decoder with Turbo-Decoding Message-Passing(TDMP)scheduling was proposed,and the decoding pipeline and some function modules were optimized.Thus the decoder has the feature of high throughput,small consumption of hardware resources and high reuse rate.3.The construction method of parity check matrix.As the performance of a LDPC code and the complexity of its encoder and decoder are all closely related to its parity check matrix structure,a method to construct a class of LDPC codes was proposed.The quasi dual-diagonal and quasi cyclic matrix structure was adopted as the fundamental structure of parity check matrix in this method.An improved Progressive Edge-Growth(PEG)construction algorithm and an optimized design method of LDPC codes with certain structure were put forward.The LDPC codes constructed in this way can be encoded quickly by using the parity check matrix directly.It can avoid the data updating conflict in layered decoders as well and obtain more coding gain.