Study On Encoding And Decoding Algorithms Of Polar Codes Based On Multi-core Processor

As the only error correcting channel codes that can be theoretically proven to achieve the channel capacity,polar codes has become an academic hotspot once it has appeared and has become the coding scheme for5 G control channels with the efforts of many people.This article starts with the basic encoding and decoding algorithms of polar codes,and then focus on the fast decoding algorithms,rate-compatible schemes and HARQ,product code encoding and decoding algorithms and the software decoders design and application of polar codes.The first chapter is the introduction,which mainly outlines the development process of polar codes and the research status of related technologies.It also introduces the common processing tools and finally shows the structure of the full article.The second chapter mainly introduces the encoding and decoding algorithms of polar codes,which lays the theoretical foundation for the further research and application of polar codes.Firstly,the basic encoding method of polar codes is given,and then the principle of channel polarization is explained in detail.Then,several common construction methods of polar codes are listed,such as the Bhattacharyya parameter method,the Monte Carlo method,and transition probability matrix method,Gaussian approximation method and standard construction method in 5G,and then show the specific implementation.Then,the SC decoding algorithm in binary tree form is discussed in detail;the principle of the SCL decoding algorithm is introduced;the working principle of the CA-SCL decoding algorithm is explained in detail,and the specific process of the algorithm is given.Finally,the performance of various decoding algorithms are compared and the advantages and disadvantages of different algorithms are discussed based on the simulations under the Gaussian channel.The third chapter mainly introduces the fast decoding algorithm and SIMD implementation of polar codes.Firstly,the principle and processing of SIMD instructions are introduced;then the SSC decoding algorithm of polar codes and the Fast-SSC algorithm based on SSC algorithm are studied,the simplification of fast algorithm is discussed,and distribution of various nodes at different code rates are also detailed.In the end,we use parallel tools such as multithreading and SIMD to implement the inter-frame software decoder of polar codes.It focuses on principle of the main thread and the modules of sub-thread,and gives the SIMD implementation of each module.Finally,the test of the implementation on the server is discussed.The performance of different simplified algorithms,the actual throughput and delay are detailed.At the same time,the impact of number of quantization bits on the performance of the decoder is analyzed.The fourth chapter mainly introduces the HARQ scheme of polar codes and its SIMD implementation.First,the two rate-compatible technologies of polar codes are briefly introduced for puncturing and shortening,and the performance of two different technologies at different code rate is analyzed.Then the principle of the HARQ scheme of polar codes is introduced,including chase combining,incremental frozen,and incremental code length.Then,the pros and cons of different technologies are compared by simulations,and the reasons are discussed.The advantages and disadvantages of different technologies in practical applications are discussed.We propose an independent decoding scheme for retransmitted frames based on the incremental code length retransmission,and we introduce the principle,and verify with the program that the performance of independent decoding is also optimal at the code rate.Finally,we implemented an incremental code length retransmission software system for polar codes.We also measured the throughput on the server,and analyzed it with the results.Chapter 5 mainly introduces the iterative decoding of the polar product codes and the SIMD implementation.Firstly,the construction principle of the classic product codes is briefly introduced;then the non-systematic polar codes is used as the component codes to construct the product codes.The principle of construction and encoding for the polar product codes are discussed.In order to reduce the complexity,we combined a regular-flip SC algorithm with a Fast-SSC algorithm to design a fast regular-flip SC algorithm,and employed the algorithm as the component decoders of product codes to implement high-speed SISO iterative decoding of product codes.The important steps are showed in the SIMD algorithm,and the performance of different decoders is discussed in combination with simulations.We also analyze the acceleration of the SISO decoding of the product codes based on the measured throughput.