Turbo码的几种译码算法及性能比较

Kinds of decoding algorithm and performance comparison of Turbo

本文主要描述了Turbo码译码的具体算法和实现的性能,几种算法是:MAP、Max-Log-MAP、Log-MAP和SOVA算法。MAP算法被用于卷积码的译码,但用作Turbo码的译码还是要做一些修改;Max-Log-MAP与Log-MAP是根据MAP算法在运算量上做了重大改进,更加适合于实际系统的运用;Viterbi算法并不适合Turbo码的译码,修改后的具有软信息输出的SOVA算法,就正好适合了Turbo码的译码。这些算法在复杂度上和性能上具有一定的差异,系统地了解这些算法的原理是对Turbo码研究的基础,同时对这些算法的复杂度和性能的比较研究也将有助于Turbo的应用研究。

It is described some algorithms and performances of Turbo decoding , refer to the algorithm, they are MAP, Max-Log-MAP, Log-MAP and SOVA . MAP is used to decode convolution code, when used to decode Turbo, must be modified. Max-Log-MAP and Log-MAP are improved greatly on operating sup according as MAP, which is could be used for practice operation. Viterbi algorithm is unfit for Turbo decoding while Viterbi is modified and change into SOVA algorithm having soft information output, then the SOVA is fit for Turbo decoding. There are some differences among these algorithms about complexity and performance. It is necessary to understand the algorithms for studying Turbo, meanwhile the comparison of complexity and performance is in favour of application study of Turbo.