Research On Raptor Code Based On High Performance QC-LDPC Precoding

The development of application services such as webcast platforms,Virtual Reality technology,and unmaned driving has put forward higher requirements for mobile communication technology.In order to meet the needs of large-scale user access,the goals of low latency,high reliability and low consumption to provide better service quality,under the circumstance that the existing channel coding scheme can no longer meet the technical requirements,a new solution is urgently needed.Raptor code has the advantages of rateless,no need for feedback and retransmission,and low coding and decoding complexity.It has been more and more studied and applied to the fields of multimedia transmission technology,picture,audio and video compression.As an error correction coding technology at the application layer,Raptor code has huge development potential in future application services.Raptor code inherits the feature of rateless of LT code,and can send data encoded packets without restrictions until the decoding is successful,which solves the problems of information loss and poor scalability in the traditional channel coding method with code rate.Raptor code has the characteristics of strong adaptive ability and strong scalability.Aiming at the problems of short loops and complicated hardware implementation in the general construction of the check matrix of LDPC（Low-Density Parity-Check Codes）code which is the outer code of Raptor code,this paper proposes a large perimeter length construction algorithm based on QC-LDPC code by combining algebraic method and search method.This algorithm not only solves the above problems,but also enhances the BER performance of Raptor code.The specific research contents are as follows:1.The encoding and decoding process of LT code and Raptor code is expounded.According to the important influencing factors such as degree distribution and decoding method,Matlab is used for performance simulation and comparison analysis.The basic ideas and construction methods of LDPC and QC-LDPC are introduced.The performance simulation comparison of eliminating short loop is carried out.2.According to the short-loop characteristics of QC-LDPC code,the method of constructing the check matrix of QC-LDPC code with long girth based on the fundamental matrix is introduced.This paper combines the advantages of the algebraic construction method with low complexity and the search construction method to obtain a check matrix with a large girth.By adding ring of integers constraints,a check matrix with lower search complexity and a smaller improvement factor is obtained,which overcomes not only the difficulty of looking for the large-girth codeword which greater than or equal to 10 due to the algebraic nature of this method,but also the shortcoming that the search space of the method is proportional toO(（z-1）^m+n-4).The simulation results show that when the SNR ratio of the large girth QC-LDPC code is 2.5d B,the bit error rate is below 10^-6,and it has better BER performance.3.The codeword obtained by the large girth construction algorithm of the QC-LDPC code proposed in this paper is used as the outer code of the Raptor code to simulate.The simulation results show that the Raptor code has an average gain of 0.4d B compared to the general construction method when the BER is 10^-5.The gain improves the BER performance of the Raptor code.In addition,the first type m=5,m=6 and the second type m=3 of fundamental matrices proposed in this paper successfully search for the constraint values that satisfy the check matrix with a girth of 12 within a given range of values.The proportions are 51.9%,24.2%and 13.4%,verifying that the construction method of adding constraint algebra combined with search is effective.