Research On Soft Decision Decoding Algorithm Of Linear Block Codes

With the further development of error correction coding techniques,soft-decision decoding algorithm attracted widespread attention among codingscholars. This paper mainly studies short binary cyclic codes based decodingalgorithm, improves the traditional ordered statistics decoding algorithm, andconstructs a new LDPC code with the existing LDPC codes’ parity check matrixand its layered decoding algorithm. Details are as follows:1. Based on the special nature of any k cyclic consecutive bits of cyclic codecan be used as information bits, a new candidate codeword generate algorithm isgiven: select the appropriate codeword segments of k cyclic continuous bits asinformation bits in the cyclic code, and then for each section, select the s leastreliable bits to generate a list of candidate codeword, finally the mostappropriate candidate codeword is selected as decoding result.2. The algorithm of complexity reduction of list decoding by using sideinformation is given, then in-depth analysis the reasons of complexity reductionand side information based on shift register sequence is provided, comparedwith the method of using stable channel to transmit side information, it does notuse additional channel, and also not affect the decoding performance andcomplexity, more easy to engineering realization. Finally improve the traditionalordered statistic decoding algorithm by using a small amount of the mostreliable bits as side information, thus significantly reduce the decoding list.3. The method of construct a new longer LDPC code’s parity check matrix witha list of shorter ones is proposed, the LDPC codes constructed by using thismethod have ample sub-code structure and linear encoding complexity, based onthe special structure of the LDPC code we give its layered BP decodingalgorithm. Simulation results show that at high signal to noise ratio layered BPreduces the decoding complexity, while the bit error rate is not inferior to theclassic BP decoding algorithm.