Research And Application Of Polar Coding And Decoding Technology

Polar codes are the only coding scheme that can achieve the channel capacity of arbitrary binaryinput discrete memoryless channel through theoretical proof.Polar codes have been written into the standard as a channel coding scheme for control channel in 5G enhanced mobile broadband(eMBB)scenario.The communication system of B5 G should have higher reliability,larger capacity and lower delay,which puts forward higher requirements for channel coding.The thesis focuses on the polar codes,studies the decoding algorithm and a new coding scheme based on polar code,and studies the application of polar codes in multiple input multiple output(MIMO)system.Firstly,the thesis introduces the basic principle and encoding and decoding methods of polar codes.By introducing the process of channel combination and channel splitting,we show the theoretical basis of polar codes-channel polarization phenomenon,and give the construction methods of the generation matrix and information bit set of polar codes.Based on the detailed discussion of the classical decoding algorithms of polar codes,including successive cancellation(SC)decoding,SC list(SCL)decoding,CRC aided SCL(CA-SCL)decoding and belief propagation(BP)decoding algorithm,the effects of different code length,code rate and construction methods on the performance of polar codes,as well as the performance and influencing factors of different decoding algorithms are studied,and the simulation results are given.Combined with the Na-BPL decoding algorithm and bit-flipping decoding algorithm,a bit-flipping noise-aided BP list(Na-BPL)decoding algorithm is proposed.According to the different error probability of different bit positions under the condition of high-order modulation,we improve the BP correction decoding algorithm and simulation results show that the decoding performance of the propose decoding algorithm is improved without increasing the average complexity.Then,as new channel coding scheme,polarization adjusted convolutional(PAC)codes which are based on polar codes are studied in the thesis.Firstly,the coding method of PAC codes is introduced,and the process of convolution transform and polar transform is described.Then,two decoding methods,sequential decoding and list decoding,are introduced.Focusing on the construction of PAC codes during coding,combined with genetic algorithm,we propose an optimization method of the data index set of rate profiling of PAC codes based on genetic algorithm,design the methods of selection,crossover and mutation of genetic algorithm,and put forward a scheme to maintain population diversity in the iterative process of genetic algorithm.The simulation results show that the PAC code constructed by the data index set that optimizized by the proposed method has better performance than the PAC code constructed by the information bit set construction method of polar codes.Further extended to the bit interleaved coded modulation(BICM)scenario,the PAC code constructed by the method proposed in the thesis still has better performance.Finally,the thesis studies the MIMO system combined with polar codes.Firstly,the MIMO detection algorithms are studied.The classical detection algorithms such as zero forcing,minimum mean square error and maximum likelihood detection(ML)are introduced.The soft interference cancellation(SoftIC)detection algorithm using interference cancellation is analyzed,and the performance of different detection algorithms in different antenna scale scenarios is simulated and analyzed.Aiming at the problem that the complexity of ML detection increases exponentially with the increase of the number of transmitting antennas,combined with the idea of parallel interference cancellation,an iterative grouping ML detection algorithm is proposed,which controls the number of antennas in each group within the acceptable complexity of ML detection by grouping.Considering polar codes as channel coding,MIMO system with independent detection and decoding and MIMO system with iterative detection and decoding are introduced.Taking SoftIc detection as the detection module of the latter,large iterative detection and decoding system based SoftIC is introduced.Simulation results show that the large iterative system has better performance than independent detection and decoding.In order to use the SCL decoder which is without soft output of polar codes in large iterative system,a method of constructing feedback soft information through multiple alternative decoding result vectors is introduced.The simulation results show that the soft information constructed by this method can drive the large iterative system to work normally,and has better performance than the independent detection and decoding system.