基于卢比变换码的分解重组编码算法

Decomposition and Restructuring Encoding Algorithm for LT Codes

LT(卢比变换)码是第一个能够实际应用的数字喷泉码。尽管采用BP(置信)译码算法的LT码的编译码复杂度已接近线性,但依然会增加较大的时延。文章提出了一种DRE(分解重组编码)算法,对所有信息包分解后重新组合,分成多组并行编码,有效地降低了编码时间和译码的操作数,减少了数据传输时延。该编码方法可适用于不同度分布函数。仿真结果表明,当信息包数量K=1 000和2 000时,采用同样的度分布可以达到NRE(非重复编码)的性能,并能够减少约75%的编码时间和大量的译码操作数。

LT codes are the first realization of the digital Fountain codes. LT codes using BP decoding algorithm which means the complexity is close to that of linear algorithm. However it still suffers from great delay. In order to solve this problem, we propose a decomposition and restructuring encoding algorithm. The algorithm first decomposes and regroups K information packets, and then divides the packets into a plurality of sets for parallel encoding. The algorithm can effectively reduce the en- coding time and operation of decoding with reduced data transmission delay. The encoding method can be applied to various modifications distribution. In the case of K = 1 000 and K = 2 000 and the same degree distribution, the simulation results show that it can achieve Non-Repetitive Encoding （NRE）algorithm performance with 75 % encoding time reduction and less decoding operands.