Sparse code multiple access(SCMA)technology was proposed by Hosenin Nikopour et al in 2013.It is a non-orthogonal multiple access technology based on code domain,and evolved from low density signature(LDS)scheme.The multi-user detection of SCMA at the receiver uses message passing algorithm(MPA).The MPA algorithm utilizes the sparsity of the SCMA codebook to reduce the complexity,but it still has a high complexity in practical applications.This thesis analyzes the decoding process of the classical MPA algorithm and finds that its complexity mainly coming from the iterative update process,i.e.,the iteratively transmitting external information between the variable node and the function node.Therefore,this thesis mainly researches the MPA algorithm to reduce the complexity in terms of the maximum number of iterations and the number of messages participating in the iterative update.The details are as follows:1.Aiming at the slow convergence speed of classical MPA algorithm,the working mode of MPA algorithm based on serial strategy and parallel strategy re spectively are compared and analyzed.It is found that the MPA algorithm in serial mode can improve the convergence speed of external information.However,the current serial MPA algorithm fixed node update order,and the decoding performance needs to be further improved.Therefore,this thesis proposes an adaptive update message passing algorithm(AU-MPA)based on external information.The algorithm uses the residual value of each iteration as the dynamic selection criterion.Firstly,the external informat ion of the node with the largest residual value is selected,and then the external information of the node independent of the node with the largest residual value is adaptively updated.The simulation shows that the decoding performance of the 2 iterations of the AU-MPA algorithm obtains only about 0.2dB performance loss compared with the 6 iterations of the classical MPA algorithm.Further at the case of similar performance requirement,the AU-MPA algorithm speeds up the convergence than other serial MPA a lgorithms.Therefore,the AU-MPA algorithm relieves decoding time burden by reducing the maximum iterations number.2.Aiming at the problem of redundant computing in the iterative update process of classical MPA algorithm,this thesis first analyzes the conditional probability density function of the received signal and finds that there are increasing numbers of conditional probability density value with the increase of SNR,however these values is nearly useless with decoding.Therefore,this thesis proposes the conditional probability selection update MPA(CPSU-MPA).The CPSU-MPA selectivity updates the conditional probability density value by setting reasonable weights to reduce the amount of message update calculation in the iterative process.Further t he CPSU-MPA algorithm adopts the compensate method for the discarding external information,which ensures the smaller algorithm cost since part of the information is discarded.The simulation results verify that the CPSU-MPA algorithm achieves the reducing decoding complexity with the acceptable cost of the bit error rate performance. |