Decoding Performance Analysis Of LDPC Codes In Deep Space Communication

Due to long distance and great signal loss of deep space communication, it is necessary to adopt high-gain channel codes and other techniques to protect the information transmission reliably and effectively. LDPC (Low-Density Parity Check) codes have been received more and more attention for their outstanding performance in deep space communication. This thesis deals with issue that how to analyze decoding performance effectively, and to design superior performance of LDPC codes. Based on analysis of basic concepts of LDPC codes and encoding algorithms, BP, min-sum and improved min-sum decoding algorithms are discussed in detail.Experience shows that the improved min-sum algorithm is closer to BP algorithm in accuracy, and less complexity in practical application.Then, several decoding performance analysis algorithms are researched. EXIT chart algorithm which is utilized to analyze performance through mutual information, is more accurate and robust than Density Evolution Theory and Gaussian approximation methods. A new APSO-EXIT chart algorithm is proposed since noise threshold and degree distributions can't be searched by EXIT chart algorithm automatically. An overall cost function is designed to measure the matching degree of EXIT curves. Then the degree distributions are optimized iteratively by Adaptive Particle Swarm Optimizer (APSO) algorithm, which doesn't need to fix the CND curve. Therefore, more matched degree distributions and higher noise threshold closer to Shannon limit could be obtained.Finally, a check matrix in good performance is constructed by means of the degree distributions optimized from APSO-EXIT chart algorithm. Two LDPC encoding/decoding schemes with LDPC-SPC algorithm are designed according to the requirement of deep space communication. Simulation results show that both schemes can reach BER less than 10-5 while SNR is above 0.7 dB, and APSO-EXIT chart algorithm is effective for decoding performance analysis.