On The Investigation And Improvement Of Low-Complexity Decoding Algorithms For Polar Codes

Polar codes has been selected as the control channel coding scheme for 5G enhanced mobile broadband(eMBB)scenarios in 2016.The CRC-Aided Successive Cancellation List(CA-SCL)decoding and the Progressive Bit-Flipping(PBF)decoding for polar codes have good error-correction performance.However,they have higher decoding complexity and delay.To reduce the decoding complexity and delay of CA-SCL and PBF algorithms,this thesis studies the distribution of error bits in the Successive Cancellation(SC)decoding.The main work is as following.Firstly,the CA-SCL decoding needs to perform path splitting for each information bit,and then filter the path according to the path metric value.This process will result in a large decoding delay.This thesis analyzes the distribution of low-reliability information bits in the Successive Cancellation(SC)decoding and construct the set of low-reliability information bits,so that the path splitting is only performed at the low-reliability information bit position in the CA-SCL decoding,which called Split-reduced CA-SCL(SR-CA-SCL)decoding.The Specific SR-CA-SCL(SSR-CA-SCL)algorithm based on the specific information bit position splitting rule and the Dynamic SR-CA-SCL(DSR-CA-SCL)algorithm based on the dynamic information bit position splitting rule are proposed.Compared with the CA-SCL algorithm,the SSR-CA-SCL and DSR-CA-SCL algorithms can significantly reduce the number of path splitting of CA-SCL decoding and without degrading performance.For polar code(1024,512),at 1.0dB,2.0dB,3.0dB,the SSR-CA-SCL decoding can reduce the number of path splitting by 61.1%,76.6%,92.4%,respectively,and the DSR-CA-SCL algorithm can reduce the number of path splitting by 69.1%,87.2%,97.5%,respectively.Secondly,the conventional PBF decoding algorithm can correct a plurality of error bits caused by channel noises by performing multiple bit-flipping decoding attempts.However,the PBF decoding is similar to a tree search process,which has high decoding delay and complexity.The size of the critical set containing the error bit positions in the PBF is positively related to the decoding delay and complexity.Therefore,we propose a more efficient method to construct a smaller set of error bit positions.In addition,to further reduce the number of SC decoding attempts for correcting multiple error bits caused by noise in PBF decoding,we propose a pruning technique for reducing the number of error bit-flipping combinations.Applying the method of constructing error bit set and the technique of pruning error bit-flipping combinations to PBF decoding,an improved PBF(IPBF)decoding algorithm can be obtained.Compared with the PBF algorithm,the IPBF algorithm not only significantly reduces the decoding complexity,but also has a slight improvement in decoding performance.For polar code(1024,512),at 2.0dB,the decoding complexity can be reduced by 63.6%,87.5%,97.6%when correcting 2,3,4 error bits caused by channel noises,respectively.