Research On The Encoding And BP Decoding Algorithms Of Polar Codes

Polar code is the first error-correction code that can reach the Shannon limit of the communication channel,and has been the focus of academic and industry.The encoding of polar code is based on the channel polarization phenomenon,and the channel with high reliability is selected.The frozen bits selection algorithm is also a research hotspot.Since polar code can also be regarded as a Tanner graph based code,the Belief Propagation(BP)decoding algorithm used in Low Density Parity Check(LDPC)is alternative,which has natural parallelism and is suitable for high throughput and low latency application scenarios.In order to obtain better BP decoding performance,a frozen bits selection scheme of polar code based on Monte Carlo simulation is proposed,which can optimize the encoding process according to the selected decoding algorithm,and we evaluate the decoding performance under this encoding scheme.The numerical results show that the proposed frozen bit selection algorithm has better performance than the original BEC channel-based algorithm,and the proposed algorithm is more general.In addition,in order to reduce the decoding latency and hardware resources of the BP decoder,we start with the Tanner graph and study the relationship between the codec algorithm of polar code and its kernel matrix.Based on the factor graph,we propose the BP decoding iterative formula based on the high-order kernel matrix,clarify the approximate calculation method of the confidence for the variable nodes,and give the improved algorithm flow.Besides,we evaluate the performance of the decoder,and analyze the decoding latency and complexity.Finally,we design a BP decoder based on the fourth-order kernel matrix.The simulation and synthesis results show that it has the performance equivalent to the original polar code,and effectively reduces the decoding delay and storage resources by half,which has a high ratio of throughput to area.