Research On Multiuser Detection Algorithm For Sparse Code Multiple Access System

Due to the rapid growth of mobile Internet and Internaet of Things,the number of mobile terminal equipments are growing and the existing spectrum resources are becoming increasingly insufficient.The traditional mobile communication systems have been severely challenged.In order to effectively increase the utilization of spectrum resources in mobile communication systems,novel multiple access technologies emerge as the times require.Sparse Code Multiple Access(SCMA)is one of the candidates for novel multiple access technology in 5G.It allocates user-specific codebooks to implement multiplexing of time-frequency resources.SCMA can support system overload and enable more device access.This thesis mainly focuses on the multi-user detection algorithms at the receiver side of the SCMA system.The specific content is as follows:The basic principle of SCMA SCMA technologyis introduced firstly.In addition,the process of codebook design at the transmitter and the principle of diverse detection algorithms at the receiver in the system are described in detail,where the performance comparison of these detection algorithms is given.The Message Passing Algorithm(MPA)can effectively use the sparse characteristics of the system to avoid the exhaustive search to reduce the complexity,but the detection complexity still increases dramatically with the increasing device connections.This thesis also summarizes the methods of reducing complexity from different perspectives.Secondly,this thesis studies the convergence performance of codeword probability for MPA algorithm.The difference between the codeword probability of the current iteration and the previous iteration is applied to represent the degree of probability convergence of codeword.Based on this,a MPA algorithm based on avoiding redundant iteration is proposed.If codeword probability of all users reaches the target convergence,the iteration will stop.According to the asynchrony of codewords probability convergence for different users,a dynamic factor graph reduction based MPA algorithm is proposed.The algorithm can dynamically remove variable nodes and function nodes that are unrelated to subsequent iterations from the factor graph.The simulation results show that the both algorithms can achieve a certain bit error rate performance gain and reduce the detection complexity.Finally,in order to improve the detection performance,an optimization algorithm based on iterative detection decoding algorithm based MPA is proposed.This algorithm can eliminate inter-user interference and obtain more system gain through the interference cancellation,and can also reduce the load rate of outer iteration.In addition,the existence of the cyclic redundancy check module not only avoids error propagation but also dynamically adjust the number of iterations to reduce the detection complexity.Simulation results show that the optimized algorithm can achieve a better error block rate performance with low detection complexity.