高误码率下归零Turbo码参数识别

Blind Identification of Turbo Codes on Trellis Termination at High Bit Error Rate

针对现有归零Turbo码码长、帧同步、码率等参数识别算法存在容错性能低且计算复杂度高的缺点,提出了一种具有高容错性的识别算法。利用每帧初始两码元的差分结果为确定零点特性,构建新的差分序列分析矩阵,通过改变分析矩阵的列数并作特定的矩阵变换,实现码长识别。基于极大极小准则,解算出确定零点与随机零点的判决门限,实现确定零点识别,然后通过遍历确定零点,完成帧同步识别。从工程实际以及编码约束关系出发,遍历码率以及寄存器个数的可能值,实现码率以及寄存器个数识别。仿真实验表明:差分确定零点位置分布与总结的3个特点一致,判决门限能够有效判断差分确定零点;该算法容错性能较强,在误码高达0.20条件下,各种参数识别率能够达到96%以上,且所需码元数据量以及计算复杂度远小于现有算法。

In order to overcome the low fault tolerance and high computational complexity of the existing algorithms for identification of code length,frame synchronization,bit rate and other parameters of Turbo codes on trellis termination,a new recognition algorithm with high fault tolerance is proposed. Because the differencing result of two initial code elements of each frame is zero,a new differential sequence analysis matrix can be constructed. The code length can be recognized by changing the columns of the analysis matrix and conducting the specific matrix transformation. The decision threshold of definitized zero and random zero is solved based on the minimax criterion,and then the frame synchronization is identified by traversing the definitized zero. According to the actual project and the coding constraints,the code rate and the number of registers can be identified by traversing their possible values. Simulated results show that the distribution of differential definitized zero positions agrees with the three summarized features,the decision threshold can effectively be used to recognize the definitized zero point,at the same time,theperformance of the algorithm in the bit error tolerance is strong,and the recognition rate of various parameters can reach more than 96% when bit error is up to 0. 20,besides,the required amount of data symbols and computation complexity is far less than the existing algorithm.