Research On Improved Decoding Algorithms For Polar Codes

Polar coding was proposed by professor Ar?kan in 2008,which is a new type of channel coding technique with low coding and decoding complexity,and polar codes are the first capacity-achieving codes ever conceived.This new technique provides a new direction for channel coding and is expected to play a vital role in future wireless communication systems.However,in short and medium codes,the performance of the Successive Cancellation(SC)decoding algorithm for polar codes is not satisfactory.Although the Successive Cancellation List(SCL)decoding improves the error correction performance by path extension,it brings a problem of high computational complexity.In this thesis,SC decoding algorithm and its improved algorithms are researched,and two improved algorithms are proposed to make a better compromise between error correction performance and computational complexity.Firstly,the effect of error propagation is researched and SC Flip(SCF)decoding algorithm is analyzed,then a Segmented Flipped SCL(SF-SCL)decoding scheme is proposed.In the SFSCL decoding,error detection on SCL decoding is performed segmentally,and the bit-flipping decoding is introduced to try to correct the errors in SCL decoding process in advance.In order to reduce the redundant computation,a termination strategy is introduced to terminate the decoding in time when the flipping decoding fails.In addition,the SF-SCL decoding realizes multi-bit flipping,which can correct multiple errors.For increasing the probability of the success of multi-bit flipping,the distribution of low-reliability channels is further researched in this thesis.Then a segmentation scheme based on Log-likelihood Ratio(LLR)is designed,and an LLR-based SF-SCL decoding algorithm is proposed to avoid the dense distribution of unreliable channels in the same segment.Simulation results show that,compared with the traditional Cyclic Redundancy Check(CRC)-aided SCL(CRC-aided SCL,CA-SCL)decoding algorithm,the LLR-based SF-SCL algorithm has better error correction performance,and when the list is long,the latter has lower computational complexity.Secondly,the Split-reduced SCL(SR-SCL)decoding is researched and improved,and a CRC-aided SR-SCL(CA-SR-SCL)decoding algorithm is proposed.The SR-SCL decoding reduces the computational complexity of SCL decoding by eliminating the path splitting on decoding nodes with high reliability.Based on the splitting rule of SR-SCL decoding,the splitting characteristics of the correct decoding path are analyzed.Then a new pruning rule is proposed for the SR-SCL decoding,in which a relative splitting number threshold is set to delete the error path in time.Furthermore,based on the pruning rule,CA-SR-SCL algorithm introduces CRC-aided path selection.Simulation results show that,the new pruning rule reduces the complexity of SR-SCL decoding,and improves the error correction performance.The CA-SR-SCL decoding algorithm further improves the correction performance of SR-SCL decoding,and the higher the SNR is,the more obvious the gain will be.