基于陷阱检测的咬尾卷积码译码算法

Trap Detection Based Decoding Algorithm for Tail-biting Convolutional Codes

该文分析了循环维特比算法(CVA)中存在的循环陷阱问题,并证明了传统基于CVA的咬尾卷积码译码算法中存在的不足,提出了一种高效率的咬尾卷积码译码算法。该算法通过检测两次不同迭代中获得的两条最大似然路径是否相同来判断是否有循环陷阱产生,并及时终止循环,减少冗余迭代;在没有循环陷阱产生的情况下,新算法比较当前迭代中最大似然路径和已经发现的最优咬尾路径是否相同来自适应终止迭代。文中对循环陷阱检测方案和自适应终止方案做了进一步优化,即利用路径的净增量而非路径本身作为检测量。实验结果表明新算法提高了译码效率,降低了译码复杂度。

There exists circular trap in Circular Viterbi Algorithm （CVA） and deficiencies in CVA-based decoding algorithms of Tail-Biting Convolutional Codes （TBCC）. A high efficient decoding algorithm is proposed for TBCC. The checking rule for circular trap in the new algorithm is that comparing whether the two maximum likelihood paths obtained from two different iterations are identical to each other, if they are identical, the CVA should be terminated. Meanwhile, when there no trap happens, a new adaptive stopping rule for CVA is proposed which is based on comparing the maximum likelihood path with the best maximum likelihood tail-biting path. Furthermore, the path used as the measurements in the checking rule and in the stopping rule is replaced by its net path metric to reduce the complexity of decoder. The results of experiments show that the new algorithm improves the decoding efficiency and reduces the decoder complexity.