(2,1,m)卷积码的简化Viterbi译码方法

Simplified Viterbi Decoding for (2,1,m) Convolutional Codes

本文提出一种新的简化Viterbi译码方法,并讨论了新方法的误码性能和译码运算量,这种方法在不牺牲太多性能增益的条件下,平均译码运算量只是经典Viterbi译码的一半,用软件或串行ACS电路实现译码器,可提高一倍多的译码速率。

In recent years,increasing attention is paid to reduce computation of Viterbi Algorithm and complexity of decoder [1-3]. In this paper a new simplified Viterbi decoding is presented. The authors take two effective measures: reduction of states of decoder and computation of multi-branch in one step. These two measures bring the following benefits: (1)reduction of computation of Viterbi Algorithm by more than 50%; (2)reduction of hardware in the case of parallel decoder; (3)increase of decoding speed by more than 100% in the case of series decoder. In the following two paragraphs,the authors' contributions will be stated and then concisely explained. The authors' method is believed to be better than that of Ichitani [1] and Li [2] in that it is more suitable for soft decision. The method of Ichitani and Li employ syndrome to discern errors and are thus not easily applied to soft decision. The authors' method is believed to be better than that of Li [2] in that its computation is fixed, but that of Li's method is not fixed. So the authors' method is easy to implement on hardware and it can effectively increase decoding speed. The authors has developed a series decoder with microcomputer. Its bit-error probability is the same as Viterbi decoder, but its decoding speed is higher than that of Viterbi decoder.