Research On Soft Output Decoding Algorithms For Polar Codes

As the youngest coding scheme in the field of channel coding,polar codes is also the only channel coding scheme that has been strictly proved to reach the channel capacity.It has been selected into the control channel coding standard of 5G mobile communication.In recent years,polar codes have attracted the attention of many scholars because of their complete theoretical support,novel coding construction methods and excellent decoding performance,and become a hot research topic in the field of coding.The existing soft output decoding algorithms of polar codes include the belief propagation(BP)decoding algorithm and the soft cancellation(SCAN)decoding algorithm.The BP decoding algorithm has the natural parallel characteristics,and has certain advantages in reducing the delay and improving the throughput performance.The SCAN decoding algorithm has the characteristics of simple hardware implementation.However,the current BP and SCAN decoding algorithms have poor BER performance and can not meet the actual communication needs.Therefore,this thesis studies the two soft output decoding algorithms: BP and SCAN of polar codes,and proposes improved algorithms based on these two soft output algorithms.The main research results include the following aspects:1.Based on the BP decoding algorithm of polar codes,a BP-FCN-BP cascade decoder is proposed for the more practical additive Gaussian colored noise(ACGN)scene by using the fully convolution networks(FCN)in deep learning field.The system model of the BP-FCNBP cascade decoder is given,and the ACGN channel with physical significance in the system model is modeled,then the FCN network in the cascade decoder is built.After training the FCN model based on the ACGN channel,the BP-FCN-BP decoder is obtained by a cascade method.Based on the ACGN channel,the error performance of the proposed BP-FCN-BP cascade decoder is tested.The simulation results show that the proposed BP-FCN-BP decoder is very helpful to improve the error performance in the ACGN channel,and the decoder has a certain universality and generalization.2.Based on the SCAN decoding algorithm of polar codes,this thesis proposes four soft cancellation iterative decoding algorithms SCAN-Flip: SCAN-Flip R,SCAN-Flip L,SCANFlip RL and SCAN-Flip-List,which are base on critical set(CS).This thesis introduces the SCAN decoding algorithm of polar codes and gives its pseudo code.At the same time,it compares SCAN with BP decoding algorithm and summarizes the similarities and differences between the two soft output decoding algorithms.Based on AWGN channel,the performance and average normalized complexity(ANC)of the four SCAN-Flip algorithms proposed in this paper are simulated.The simulation results show that the four SCAN-Flip algorithms have a certain gain compared with the traditional SCAN algorithm,and with the increase of signal-to-noise ratio,the improvement of error performance is more obvious,but the ANC is closer to the traditional SCAN decoding algorithm.The simulation results show that the SCAN-Flip algorithms have a bright prospect in high SNR.