Research On The Lossless Simplified Successive Cancellation Decoding Method Of Polar Codes

Polar code is the first and only channel coding scheme that is proven to achieve the capacity of the binary symmetric discrete memory-less channel.It is also been adopted as the control channel coding scheme of the fifth-generation mobile communication system’s enhanced mobile broadband scenarios.The construction,encoding,and decoding of the polar code are concise.And it shows sustained and excellent reliable performance as long-range code length and rate,which makes it a hot solution for the basic technology of the physical layer of the sixth-generation mobile communication system under research.As the first decoding algorithm of polar codes,successive cancellation(SC)decoding’s performance is far from the Shannon limit on the condition of limited code length.Then,successive cancellation list(SCL)decoding is proposed to improve the SC which brings better performance.But they both suffocate from their serial structure of the decoding procedure,which drag down their decoding latency and throughput performance.As effective means to reduce the decoding latency of polar codes and improve the decoding throughput,the simplified decoding methods of polar code SC decoding and its enhanced algorithms drawn a lot attention during these years.This thesis studies the polar code’s simplified SC decoding scheme,the parallel architecture,and serial software decoding schemes of simplified SCL decoding respectively.Firstly,the simplified SC decoding algorithm for polar codes is studied.It has the characteristics of low complexity and computing resource requirements.The research on the simplified SC decoding algorithm of mixed-multikernel polar codes can further enhance its practicability in meeting the requirements of various code lengths However,due to the complexity of the composition of the polar kernel of mixed-multikernel polar codes,the research on the related simplified decoding algorithms is also slow.In this thesis,a column deduplication and simplification method for generation matrix of linearly polarized kernel mixed-multikernel polar codes is proposed,four generalized constituent nodes and their decoding methods are given,and the required time step is analyzed.The simulation results show that the proposed method can realize the simplified decoding of linearly polarized-kernel mixed-multikernel polar codes SC without loss of error correction performance.Secondly,the simplified SCL decoding scheme of polar codes under parallel processing architecture is studied.In order to improve the reliability of short and medium code decoding and reduce the decoding latency,the ordered error patterns’ relationship is used to reflect the reliability order of the estimated codewords.And the estimated codewords are directly generated from the minimum set of ordered error patterns.A simplified SCL decoding scheme based on minimum set is designed,and the simplified SCL decoding scheme is extended to mixed-multikernel polar codes with specific polar kernels.This paper analyzes the time step overhead of the proposed scheme compared with the traditional scheme,and verifies the protection of the error correction ability of the original SCL decoding algorithm by the proposed simplified scheme through simulation experiments.Finally,the simplified SCL decoding scheme of polar codes under serial processing architecture is studied.Using the reliability order between the ordered error patterns in the minimum set of ordered error patterns,an ordered tree data structure is constructed that can efficiently access and calculate the ordered error pattern,and two improved tournament sorting algorithms are proposed for on-demand calling.The specific computational overhead of the proposed ordered tree based single parity check(OT-SPC)software decoder in a given component node is analyzed,and decoding latency is counted.The effects of the decoder input sequence length and decoding list size on the error correction performance and throughput of related decoding methods are studied.In summary,starting from the lossless and efficient decoding requirements of polar codes,this thesis conducts an in-depth study on the simplified SC and SCL decoding schemes of polar codes.Four generalized constituent nodes for linearly polarized kernel mixed-multikernel polar codes are proposed accordingly.The SCL decoding method based on the minimum set of ordered error patterns and the SPC node decoding method based on ordered tree and improved tournament sorting algorithms are respectively given.These research results can be directly used in the simplified decoding scheme of polar codes in various scenarios,and as the basic technology research of other enhanced decoding algorithms to improve the effectiveness and reliability of the physical layer communication technology.