Performance Analysis of Maximum-Likelihood Decoding of Polar Codes
Xiangping Zheng,Xinyuanmeng Yao,Xiao Ma
DOI: https://doi.org/10.1007/978-981-97-1568-8_10
2024-01-01
Abstract:In space communication, effective channel coding schemes and decoding algorithms are essential for reliable communication. As the standard coding scheme for 5G enhanced Mobile Broadband (eMBB) control channels, polar codes are usually decoded by the successive cancellation list (SCL) decoding algorithm. As the list size increases, the SCL decoding performance is improved and approaches the Maximum Likelihood (ML) decoding performance, while leading to significantly increased complexity. Therefore, analyzing the ML decoding performance is helpful for us to select a suitable list size and hence to trade off the performance and the complexity of the SCL decoding of polar codes. In this paper, we employ the improved union bound and the lower bounding technique based on Bonferroni inequality to evaluate the ML decoding performance of polar codes. The former requires only a truncated weight spectrum and the latter relies only on a subset of the codebook. To calculate the weight spectrum of a polar code, random interleavers are introduced. In contrast, a subset of the codebook can be obtained by performing the SCL decoding algorithm. Simulation results show that the proposed bounding techniques can effectively predict the ML decoding performance of the polar codes and provide guidelines on the choice of the parameters of the SCL decoding. Specifically, the SCL decoding with list size 2 is sufficient to approach the ML decoding performance of the 5G polar codes [128, 64], [256, 128], and [512, 256]. By contrast, for the 5G polar code [128, 16] with a 4-bit cyclic redundancy check (CRC), the list size is required to be 8 to obtain near ML decoding performance.