一种采用非均匀量化的近似Log-MAP算法  

A Non-uniform Quantizing Approximation Log-MAP Algorithm

作　　者：周继宇[1] 张雅奇[1] 徐伯庆[1] 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《无线电通信技术》2012年第1期29-31,共3页Radio Communications Technology

摘　　要：为了简化Log-MAP算法、减小其硬件查表实现时所占用的外存储单元,采用了非均匀量化的方法分别对Log-MAP算法中的校正函数进行两电平和五电平近似,从而得到相应的量化函数。并将该量化函数用于LTE Turbo译码器,在AWGN信道上进行软迭代译码仿真。仿真结果表明,两电平和五电平非均匀量化的近似Log-MAP算法将Max-Log-MAP算法的译码性能提高了0.2～0.3 dB,与原Log-MAP算法性能接近,但却能大大减少查表时需要检索的数据量。This paper adopts two-level and five-level non-uniform quantization to approximate the correction function of Log-MAP so as to simplify Log-MAP algorithm and diminish the storage of hardware implementation. Then, both of them are used in turbo decoder of LTE system and simulated on AWGN channel. The simulation results show that error-correcting capability of non-uniform quantizing approximation Log-MAP algorithm is higher 0.2 -0. 3 dB at least than Max-log-MAP algorithm and near to Log-MAP algorithm, while it has better performance in hardware implementation compared with Log-MAP algorithm.

关 键 词：TURBO译码 Log—MAP算法 校正函数 非均匀量化 软迭代译码 

分 类 号：TN911.2[电子电信—通信与信息系统]