Research On Polar Code Decoding Algorithm

Since the advent of "Mathematical Theory of Communication",scholars in countless channel coding circles have approached the channel limit(Shannon capacity)in the direction given by Shannon.Although Turbo and LDPC(Low Density Parity Check,LDPC)achieve excellent performance close to the channel limit,there is no strict proof of accessibility.The polar code is the first channel coding method rigorously proven to achieve channel limit by Erdal Arikan in 2009.It has received unprecedented attention in the academic world owing to its superior low coding complexity and channel capacity accessibility.This paper focuses on and studies the theory of polar coding.It focuses on the design and research of the Successive Cancellation Decoding(SCD)algorithm and its improved decoding algorithm,and proposes an improvement segmentation-based list decoding.This article mainly includes the following two aspects of work:First,the coding principle of the polar code is analyzed.By analyzing the mutual information distribution of Binary-input Symmetric Memoryless Channels(BSMC),it is concluded that the mutual information distribution of the Binary Erasure Channel(BEC)is concentrated at the extreme value,and then the polarization principle is derived.The mutual information distribution of serial combining and parallel combining is analyzed for BSMC.Finally,the channel reliability parameters of the polar code are studied in detail.Secondly,the decoding algorithm of polar code is analyzed and an improved segmentation list decoding algorithm is proposed.Based on the research of Successive Cancellation Decoding(SCD),this paper analyzes the improved algorithms of SCD,such as CA-SCL,SSCD,SCAN,etc.The simulation results show that the list decoding algorithm can achieve maximum likelihood decoding performance,SSCD,SCAN have both efficiency and performance advantages.Then an improved segmentation list decoding algorithm is proposed: by segmenting the information bits randomly,and then encoding the segmented information blocks using Cyclic redundancy check(CRC)of different lengths.Polar coding is sent to the Additive White Gaussian Noise Channel(AWGNC)for BPSK modulation.After the list decoding by the receiving end,the corresponding CRC generating polynomial is used for checking and screening for different information blocks.In the process of performing CRC check,when the corresponding CRC check succeeds,the information path is in all information.If there is a maximum log likelihood ratio in the path,the information path is filtered and stored for the next segment,otherwise the decoding fails.The simulation results show that the proposed scheme reduces the computational unit and improves the decoding performance of the polar code without changing the complexity.It is especially suitable for communication scenarios with short code length.