Analysis And Research On Optimization Of Successive Cancellation Decoding Algorithms Of Polar Codes

Polar codes were proposed by Professor Arian in 2009.With the special coding method based on channel polar phenomenon,polar codes has become the only channel coding scheme which has been proved theoretically to reach the Shannon limit,and attracting the attention of experts in the field of channel coding.With the unremitting efforts and exploration of experts,polar codes have developed into an excellent coding scheme with excellent performance and low hardware implementation complexity.At present,polar codes have been selected as the control channel coding scheme for enhanced mobile broadband scenarios in 5G(5th-Generation)communication standard.This paper mainly studies and optimizes the SC(Successive Cancellation)decoding algorithm of polar codes.The main contents and innovative achievements of this paper is as follow:1.Based on the polar phenomenon of polar codes,the channel reliability estimation of polar codes,the encoding principle of polar codes are studied.At the same time,the SC decoding algorithm and BP(Belief Propagation)decoding algorithm of polar codes are introduced in detail.Then,based on the shortcomings of SC algorithm,the improved SC algorithm SCL(Successive Cancellation List)is deeply explored,and several improved SCL algorithms are analyzed.Finally,a joint simulation system based on Matlab and Pycharm software is built,and the performance of polar decoder based on deep neural network is deeply explored,which paves the way for the improvement of SC algorithm based on neural network.2.A simplified adaptive SCL decoding algorithm SAD-SCL(Simplified Adaptive Successive Cancellation List)based on CRC(Cyclic Redundancy Check)is proposed.SAD-SCL algorithm draws lessons from the idea of segmented CRC of SCAD-SCL(Segmented-CRC Adaptive SCL).Using new methods to reduce redundancy in AD-SCL(Adaptive Successive Cancellation List)algorithm.Before each update value decoding,SAD-SCL first uses the existing LLR(Log-Likelihood Ratio)information to determine the range of burst error bits,and then only re-decodes the information sequence of the range.Moreover,when SAD-SCL can not get the correct decoding value of the first half,SAD-SCL uses SC decoding with less computation to complete the decoding of the second half,and the result is almost the same as that using AD-SCL algorithm for the latter half of the decoding.This algorithm has lower computational complexity than AD-SCL and SCAD-SCL without loss of performance.3.A SSC decoding algorithm MIO-NSSC(Multiple In One Neural Simple Successive Cancellation)based on multi-in-one neural network is proposed.Base on the combination of SC algorithm and deep neural network in NSC(Neural Successive Cancellation)algorithm,MIO-NSSC algorithm uses parallel computing method of special nodes in SSC decoding to further reduce the decoding delay of polarization codes.At the same time,MIO-NSSC trains a more applicable deep neural network MIO-NN(Multiple In One Neural Network)to replace multiple neural networks in NSC through a new strategy,which greatly reduces the space complexity of the decoding algorithm.It has been proved that when the code length is 128 and the code rate is 0.5,the delay of MIO-NSSC decoding algorithm proposed in this paper is reduced by about 21% compared with NSC,and about 80% of the nodes is saved.