摘要
针对目前标准非删余极化码盲识别存在的算法复杂、误码适应能力不足等问题,首先证明了能够表征极化码码长、码率关系,区分信息子信道和冻结子信道的定理与命题,并基于此提出了一种高效的盲识别算法。该算法只需设定可能的最大码长,构建对应的软判决码字矩阵与克罗内克矩阵,利用所证明的定理与命题对两矩阵校验关系进行判决,估计码率及冻结比特置,最终完成参数的识别。在校验关系判决中,引入平均似然差作为检验量,基于其理论概率分布与极小化极大准则确定判决门限。仿真结果表明,推导的定理、命题与仿真结果一致,当信噪比为6 dB时,对于码长为1 024的极化码,其参数识别率仍接近100%,识别性能和计算复杂度均优于现有软判决算法。
In order to solve the problems of complexity and poor error adaptability in the blind recognition of standard non-punctured polarization codes, the theorems and propositions that could characterize the relationship between the code length and code rate, and distinguish information subchannel and frozen subchannel were proved. Based on the theorem and proposition, an efficient blind recognition algorithm was proposed. The proposed algorithm only needed to set the possible maximum code length, and the corresponding soft decision codewords matrixes and Kronecker matrixes were constructed. Based on the theorem and proposition proved, the check relationship between the two matrices was judged and the code rate and frozen bit position were estimated. The average likelihood difference was introduced as the test quantity, and the decision threshold was determined based on its theoretical probability distribution and minimax criterion.The simulation results show that the deduced theorem and propositions are consistent with the simulation results. Under the signal-to-noise ratio of 6 dB and code length of 1 024, the parameter recognition rate is still close to 100%. The recognition performance and computational complexity are better than the existing soft decision algorithms.
作者
王垚
王翔
杨国东
黄知涛
WANG Yao;WANG Xiang;YANG Guodong;HUANG Zhitao(College of Electronic Science and Technology,National University of Defense Technology,Changsha 410073,China;Communication Sergeant School,Army Engineering University,Chongqing 400036,China;Unit 92001 of PLA,Qingdao 266023,China)
出处
《通信学报》
EI
CSCD
北大核心
2022年第2期22-33,共12页
Journal on Communications
基金
湖南省自然科学基金创新群体资助项目(No.2019JJ10004)。
关键词
信道编码
标准极化码
非删余
识别
channel code
standard polar code
non-punctured
recognition
