一种简化的Turbo码并行译码方法.pdf

A Simplified Parallel Decoding Scheme for Turbo Codes Wang Le, Huang Yu, Yang Hongwen Wireless Communication Center, Beijing University of Posts and Telecommunications, Beijing, China (100876) E-mail ：mekingfun@ Abstract Parallel decoding of Turbo codes is vital to the applications with very high data rates. In the overlapping parallel decoding scheme, the decoding throughput can not increase linearly with the number of segments. A scheme where the throughput can increase linearly was proposed, where the boundary information of each iteration is stored to be the initial condition in next iteration. In this paper, we have simplified the scheme by only storing the index of the most probable state of the boundary. With the cost of some small performance loss, the scheme can reduce largely the extra memory required to store the boundary information. Keywords: Turbo codes; MAP algorithm; parallel decoding 1. Introduction With the ever-increasing demand of the data rate, turbo code is facing stringent challenges. ITU suggests that the data rate of 4G system be at least 100Mbps, in mobile conditions. As the local operating clock frequency in hardware system is typically within 100MHz nowadays, it is reported [1] that the throughput of only 6.17Mbps can be achieved for turbo codes with conventional serial decoding [2]. Obviously, the decoding algorithm of turbo code has to be improved significantly to survive in the future communication systems. The bottle neck of the throughput lies in its decoding scheme. The convolutional codes is t