基于可译集大小的LT码编码算法的改进

在中、高码率情况下,使用传统随机图法(RG)构造中短码长LT码,其生成矩阵并非一定满秩,所以存在着一定的译码失败率(信源不完全可译).针对随机图法的这种缺点,提出一种改进的累积边增加法(IPEG),此方法通过控制可译集大小,保证全部信源符号的可译性.实验结果表明,与传统方法相比,经过改进的LT码性能得到了显著提高,而且每帧中错误的码字数量也更加稳定,以这种稳定的LT码作为内码,Raptor码的性能也能得到极大提升.

Improvement of LT Encoding Algorithm Based on Ripple Size

Generator matrix of moderate and high-length LT codes,constructed by typical random graphs(RG),is not be sure full rank based on moderate and high rate conditions.There is a certain decoding failure probability(not all symbols can be recovered).To solve the problem of RG,a new construction method(named improved PEG,IPEG) for LT codes is proposed.It enables all source symbols to be recovered by controlling the ripple size(RS).Results show that compared with PEG,the performance of LT codes constructed by IPEG...