Polar码SC译码算法的量化问题  

The Quantification Problems for Polar Code with Successive Cancellation Decoder Algorithm

作　　者：樊婷婷[1] 杨维[1] 许昌龙[1] 

机构地区：[1]北京交通大学电子信息工程学院,北京100044

出　　处：《北京邮电大学学报》2015年第4期95-100,共6页Journal of Beijing University of Posts and Telecommunications

基　　金：国家自然科学基金项目(51274018);国家"十二五"科技支撑计划课题(2013BAK06B03)

摘　　要：为了在数字芯片上以低硬件复杂度实现Polar码的译码算法,对Polar码连续消除(SC)译码算法接收符号和SC译码输入的初始比特对数似然比(LLR)的量化问题进行了研究.分析了接收符号量化区间和量化比特数对Polar码SC译码性能的影响.对译码输入初始LLR,从均匀量化和非均匀量化两方面,并对非均匀量化采用了归一化非均匀量化和小数非均匀量化2种方式,分析了初始LLR的量化对Polar码SC译码性能的影响.仿真结果表明,分别对接收符号和初始LLR采用区间[-4,4]和区间[-20,20]上的6 bit均匀量化,就可以使Polar码SC译码算法的误比特率(BER)性能损失在小于0.1 d B的同时,具有更简单的硬件实现复杂度.In order to apply decoding algorithms of polar code in digital chips with low hardware complexity,the quantification problems on the received signals after channel transmission and the initial log likelihood ratios（ LLR） of successive cancellation（ SC） decoder were studied. The quantized interval and quantization accuracy for the received signals on system bit error rate（ BER） performance were analyzed.For the quantization of initial LLR,both uniform and non-uniform quantization which contain two methods named normalizing quantization and decimal quantization were applied. Simulations show that 6 bit uniform quantization on interval [- 4,4] and [- 20,20] for received signals and initial LLR respectively can both bring BER performance loss less than 0. 1 d B,but with hardware implementation complexity decreased significantly.

关 键 词：Polar码 连续消除译码 量化 误比特率 

分 类 号：TN919[电子电信—通信与信息系统]