基于两种信息融合的turbo码帧同步器研究

Frame Synchronizer for turbo Code Based on Two Kinds of Information's Fusion

提出了一种基于两种信息融合的turbo码帧同步器。对turbo码1步迭代外信息统计发现,帧同步与帧失步时turbo码的外信息的统计特性存在明显的差异。在信噪比极低的情况下,传统的基于同步码相关的方法已不能满足turbo码强译码能力的要求。为了进一步降低帧同步差错率,同时又不给系统带来太大计算延迟,对turbo译码器1次迭代后的最大后验信息和同步码字相关后的最大似然信息融合成一个置信度量,仿真表明:只需很小的额外计算量,这种方法能够获得比传统方法更低的帧同步差错率。

A new frame synchronizer for turbo code based on the fusion of two kinds of information is proposed in this paper. It is found that the statistic characteristics of the mean of the extrinsic information＇ s absolute values after the first iterative step in the case of perfect frame synchronization are distinctly different from those that are obtained when the synchronization loses. In the environment of low signal to noise ratio, the traditional synchronization schemes based on the correlation method can not match the strong error correcting ability of turbo code. In order to lower the frame synchronization error rate （FSER） while not introducing so much calculated load, the maximum a posteriori （MAP） information from turbo decoder after the first iterative decoding step and the maximum likelihood （ML） information from the ASM correlation are fused into one belief metric. Simulation results show that the novel algorithm can get much lower FSER at the price of a little more additional calculation.