一种BICM-ID内嵌Turbo码译码分级选择停止准则  

Class-Selection Stop Criterion for Iterative Decoding of BICM-ID Embedded Turbo Codes

作　　者：韩婷婷[1] 李建广[1] 

机构地区：[1]中国传媒大学信息工程学院,北京100024

出　　处：《中国传媒大学学报（自然科学版）》2014年第1期51-56,共6页Journal of Communication University of China：Science and Technology

摘　　要：本文提出一种比特交织编码调制加迭代译码(BICM-ID)内嵌的Turbo码译码分级选择停止准则。该分级选择停止准则按照信息比特的后验对数似然比(LLR)绝对值大小对信息比特进行不同可靠性级别分类,针对不同类别设置相应收敛测试准则。较可靠的类别通常更早停止迭代,以节省计算量,而可靠性较低的类别则进行更多次的迭代以获得更高的译码精准。对于较早收敛的类别,关于这些信息比特的计算几乎停止,除了对前向度量和后向度量构成的马尔科夫链的更新。由于对早期收敛的类别的不必要的计算被省略,本分级选择停止准则节省了后期迭代的计算量。结果显示,本停止准则比典型的整帧交叉熵(CE)停止准则节省将近10%的迭代处理时间,而BER性能则接近。Proposed is a class -selection stop criterion for iterative decoding of bit -interleaved coded modulation with iterative decoding （ BICM-ID） embedded turbo codes,achieved with a series of differ-ent kind convergence test criteria which are in allusion to classes with different reliability levels according to the absolute magnitude of log-likelihood ratio （ LLR） values. The early converged classes would not be computed in later iterations except the Markov chains of the forward metric and the backward metric before the iterative decoding reaches a total stop threshold. The computation of later iterations is simplified since the unnecessary computation of early converged classes is left out. Simulation results reveal that the proposed stop criterion is about 10% faster than the typical frame -based cross -entropy （ CE ） stop scheme with almost the same bit error rate （ BER） performance.

关 键 词：TURBO码 停止准则 交叉熵 分级选择 

分 类 号：TN911.22[电子电信—通信与信息系统]