Research On Key Technologies Of Construction And Decoding For LDPC Codes In Deep-space Communication

In Recent Years, the Deep Space exploration is developed frequently,low-density parity-check (LDPC) codes are standardized by Jet PropulsionLaboratory (JPL) and the Consultative Committee for Space Data Systems(CCSDS), due to the hight capacity data communication in Deep Spaceand the requirement of network coverage are increasing. As one of keytechnique to improve communication quality, LDPC codes have receivedmuch attention in academia and engineer now.Though LDPC codes can get good performance, the performance andthe application of LDPC codes are still restricted by these problems. suchas error floor, the contradiction of performance and complexity indecoding algorithm, the requirement of low bit-error ratio and lowcomplexity of hard equipment in Deep Space communication. This papermainly research on the construction and decoding algorithm of LDPCcodes and we analyze the root of these problems refered below. Improvements on construction and decoding algorithm are proposed toremission these problems. The work on construction and decodingalgorithms as follows:Firstly, we proposed the improvement on two kinds of balanceincomplete block designs (BIBD) by constructing different blocks.Different checking matrixs can be composed by those blocks.The modifiedLDPC codes are cyclic， Because improved LDPC codes have differentmatrixs, so it can gets variable rate. Results show the error floor is reduced,the performance of bit error rate better than BIBD at the same rates and inthe low rate the performance not bad.Secondly, the thesis presents a modified method on Min-Sumdecoding (MS) to pursue good tradeoff between performance andcomplexity. We reduce the error between theoretical value and practicalvalue by adding flexible correcting factor.The modified method increasesparty of multiplication and addition operators, but the complexity still low.The performance of bit error rate and convergence better thanMS. In low signal-noise ratio, the performance of bit error rate higher than normalized min-sum algorithm, but it convergence more rapidly.