一种咬尾双二进制Turbo码并行译码方案

A Parallel Decoding Scheme for Tail-biting Duo-binary Turbo Codes

针对双二进制Turbo译码使用并行、滑动窗联合译码技术时,其咬尾的编码构造和窗分割导致边界状态值难以获取的问题,提出了一种新咬尾Turbo码并行、滑动窗译码方案——扩展交叠方案。该方案采用了边界状态盲估计和滑动窗状态回溯两种新译码技术。相比于传统的边界状态度量传播方法(又称迭代法),新方法一方面提高了边界状态度量的准确性,从而加快了译码收敛速度,一定程度上减小了高信噪比下的性能损失;另一方面避免了存储前一次译码的迭代度量值,更有利于硬件设计。仿真表明,新方案在64左右的中等窗长下即可消除并行和滑动窗影响,逼近原始无并行无滑动窗译码的性能,且窗长越小,其相较传统迭代法带来的译码性能增益就越明显。该方案具有较好的实用性和应用价值,可以满足5G的高速率、低时延和低存储的数据传输要求。

When the tail -biting duo -binary Turbo codes use parallel and sliding window joint decodingtechnology,the tail-biting encoding structure and the window segmentation lead to the difficulty of obtai-ning the boundary state metric. To solve the problem,this paper presents a novel parallel and sliding win-dow decoding scheme called extended overlap scheme. The new scheme utilizes two new decoding tech-niques：the boundary state blind estimation technique and the sliding window state backtracking technique.Compared with the traditional Boundary State Metric Propagation（BSMP） method（also called the iterativemethod）,on the one hand,the new method improves the accuracy of the boundary state metric,thus speed-ing up the decoding convergence,and reducing the performance loss at a high signal-to-noise ratio（SNR）to some extent. On the other hand,it helps to avoid storing the state metric of the last iteration,which ismore conducive to the hardware design. Simulation results show that this method can eliminate the effectsof parallel and sliding window,approximating the performance of the original non-parallel and non-slidingwindow decoding method at the medium window length about 64. And the smaller the window length is,themore obvious the decoding performance gain is than the traditional iteration method. Therefore,this schemehas better practicability and application value,which satisfies the current high speed,low latency and lowstorage data transmission requirements of 5G.