Improved Decoding Of Ldpc Codes Based On Ir_harq System

A special type of LDPC codes, named structured Rate Compatible LDPC (RC_LDPC) code with flexible code rate and code sizes is proposed. Based on the essential feature of this kind of LDPC code, its Incremental Redundancy Hybrid ARQ is also presented. It reduces code rate by puncturing message bits or extending check bits. As only a part of check bits are retransmitted during retransmission, this scheme could provide better throughput performance than that of other HARQ schemes. However, this scheme requires that the FEC is Rate Compatible codes. The focus of people's research has been on how to build the RC codes, and the decoding scheme is the area to be ignored.This paper proposes an improved decoding method based on IR_HARQ scheme that uses rate-compatible LDPC codes as the FEC codes.3σprinciple indicates that if a variable X is normally distributed, 99.7% of the observations fall within 3 standard deviations of the mean, that is, betweenμ? 3σandμ+ 3σ. So it is possible to get the numbers of error bits both before and after the decoding process with the bit-error-rate (BER) correspondingly. And the computing result shows that there are always parts of bits has been decoded correctly during a decoding process despite of whether the decoding is successful. How to make good use of the former decoding results is important to enhance the performance of retransmission decoding.This paper proposes implementation of this improved decoding method. Instead of the initial reliabilities of the first decoding process, the final reliabilities from the previous decoding process could be selected as the initial reliabilities of the previously received bits during the latest decoding process. And the new decoding method calculates the privilege factor by the properties of the channel. Then the former decoding result could be used effectively. As most of error bits have been corrected during last decoding process, the new decoding method gives compensation to the distortion which the channel brings to. Moreover, the proposed scheme will not increase the complexity of the decoder.Choosing both the (2304, 1920) and (2304, 1728) LDPC codes defined in the 802.16 standard as the FEC codes, this paper simulates the performance of new and traditional decoding methods under AWGN channel with BPSK modulation and stop-and-wait scheme. And the results show that nearly 2%~20% enhance of throughput efficiency could be got while SNR is comparatively lower with the new decoding method. And during the area where SNR is comparatively higher, the enhance of throughput efficiency is not apparent, however, the decrease of iterative times reaches to 5%~20%, and that means the same trend of system decoding time-delay.