关键词: 数据分发，喷泉码，LT'码，BPML，停止集

Abstract: For Belief Propagation(BP) decoding algorithm of LT codes,stopping set prohibits the improvement of decoding efficiency. This paper analyzed and simulated the size of stopping set. A Belief Propagation-Maximum Likelihood decoding algorithm(BPML)was proposed. BPML uses BP algorithm to decode firstly. When stopping set makes BP stop, Maximum Likelihood(ML) decoding algorithm is used to deal with the stopping set. It can overcome the negative influence of stopping set and improve the decoding efficiency of LT codes. The simulation showed that BPML combines the advantages of BP algorithm in low decoding complexity and ML algorithm in high decoding efficiency. hhe conclusion of research is practically valuable in improving efficiencies of data distribution applications in computer networks.

Key words: Data distribution, Digital fountain, LT codes,BPML, Stopping set