具有检错容错能力的LT码编译码算法  

LT coding and decoding algorithm with error detection and tolerance ability

作　　者：邢艳玲[1] 裴玉奎[1] 葛宁[1] 

机构地区：[1]清华大学电子工程系微波与数字通信国家重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2013年第7期931-935,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金重点项目(61132002,61021001)

摘　　要：针对恶劣环境下接收数据中存在错误而导致喷泉码译码时的误码扩散问题,提出一种具有检错容错能力的LT(Luby transform)码编译码算法。利用LT码编码特点和BP(belief propagation)译码性质,通过对译出源包的冗余校验检出错误的编码包,结合LT码的纠删特性进行正确译码。通过增加少量译码开销,提高了LT码在接收数据错误时的传输可靠性。LT码作为Raptor码的内码,其容错设计可以作为Raptor码容错设计的参考。仿真结果表明:在采用优化度分布函数,包长为250 B,源包个数为512时,该算法可以检出任意数目编码包的错误,检错率达到99.9%以上,并且成功译码时新增译码开销不大于2%。A Luby transform （LT） coding and decoding algorithm with error detection and tolerance ability was developed to reduce error propagation caused by errors in the received data during fountain code decoding. The algorithm detects errors in the coding symbols from the checksum of the decoded source symbols using the characteristics of the LT coding and the property of belief propagation （BP） decoding. The decoding is then successfully combined with the erasure characteristic of the LT code. The algorithm improves the transmission reliability with errors in the received data with a small increase in the decoding overhead. As inner-code in the Raptor code, the error-tolerant design can be used as a reference for the Raptor code. Simulations show that this algorithm can check arbitrary numbers of error encoding symbols and achieves over 99.9% error detection with a symbol length of 250 B, 512 blocks, and an increased decoding overhead of no more than 2%.

关 键 词：喷泉码 LT(Lubytransform)码 生成矩阵 误码扩散 

分 类 号：TN911.22[电子电信—通信与信息系统]