一种低功耗的Turbo码译码算法  

Low Power Consumption Turbo Decoding Algorithm

作　　者：冯芒[1] 阎鸿森[1] 

机构地区：[1]西安交通大学电子与信息工程学院,西安710049

出　　处：《西安交通大学学报》2004年第10期1081-1084,1089,共5页Journal of Xi'an Jiaotong University

摘　　要：针对Turbo码译码器功耗大的问题,改进了传统的最大后验概率译码算法,提出了一种基于网格图合并思想的低功耗Turbo码译码算法.该算法通过减少计算过程中占功耗绝大部分的存储器的访问次数来达到降低译码功耗的目的.依据N步合并后的编码网格图进行计算,使得一次译码中计算的总时刻数变为传统算法的1/N,从而使译码器总的存储器访问次数变为原来的1/N,很好地降低了译码器的功耗.理论分析和仿真结果表明,新算法的正确性和可靠性与传统的译码算法相同,并且硬件实现中的译码时延没有增加,是一种有效、可行的低功耗译码算法.To lower power consumption in Turbo decoding, a new algorithm based on trellis combination is proposed. Given an integer combination parameter N, the number of decoding steps in the new algorithm is reduced to 1/N of conventional methods, which makes the required number of memory accesses reduced by N times. As memory accesses constitute the main source of power consumption in the decoding process, this new algorithm significantly reduces the required power consumption. Based on theoretical analysis and algorithm simulation, this new algorithm is compared in detail with the conventional maximum a posteriori algorithm from the perspectives of decoding reliability, operating process, computation complexity and practicability. The results show that the new algorithm is both feasible and effective.

关 键 词：TURBO码译码 最大后验概率译码算法 网格图 

分 类 号：TN911.22[电子电信—通信与信息系统]