Research And Implementation On Lowcomplexity Receiver For SCMA Uplink

In recent years,with the rapid development of mobile communication technology,the communication between people has become more convenient.The 5th Generation of Mobile Communication System(5G)faces higher spectral efficiency and massive device connectivity,which is a great challenge to 5G technology.Multiple access in mobile communications is divided into orthogonal multiple access and non-orthogonal multiple access.Orthogonal access is characterized by orthogonal resource allocation,and the receiver complexity is relatively low,but the number of the connection is limited.In order to meet the requirements of future communication,non-orthogonal multiple access can reuse information of different users on the same time-frequency resource.Sparse Code Multiple Access(SCMA)is a codebook-based non-orthogonal access technology,which can effectively improve the spectrum utilization and the number of user connections.However,the detection algorithm of the current SCMA system is highly complex and cannot be directly applied to practical system.Therefore,the purpose of this thesis is to reduce the complexity of the receiver and perform fixed-point implementation.The main research contents are as follows:1.In order to solve the problem of high complexity of the current detection algorithm,this thesis analyzes the existing detection algorithms,and proposes two new dynamic detection algorithms,DS-MPA and DRC-MPA.The main idea of the DS-MPA algorithm is to determine the factor graph used in the next iteration according to the degree of convergence.Based on the convergence characteristics of Message Passing Algorithms,the residual value calculation method is used as the judgement of convergence degree.The resulting factor graph is a deleted version of the original factor graph.The detection algorithm can perform signal detection according to the generated factor graph,and the calculation complexity of the algorithm is reduced by about 1/3 under the guaranteed performance.Aiming at the problem of too many ergodic combinations,the DRC-MPA solution is proposed,which removes unreliable codewords in the iterative process and reduces the numbers of constellation combination points,thus achieving the purpose of reducing complexity.The DS-MPA and DRC-MPA can be used in combination,the complexity of the algorithm can be reduced even more than 50% under the guaranteed performance.2.In order to facilitate the implementation,MPA algorithm is improved in the logarithmic domain and the fixed-point simulation is showed in the thesis.The improved algorithms include LOG-MPA and MAX-LOG-MPA.but LOG-MPA cannot effectively solve the logarithmic domain problem;while MAX-LOG-MPA can completely solve the logarithm problem,there is a serious performance loss.Therefore,a new fixed-point implementation is proposed that flexibly compensates for the loss of MAX-LOG-MPA.The simulation results show that the BER performance is improved effectively without increasing the complexity.In addition,the data that needs to be stored and quantized in the process is statistically analyzed,achieving the balance between performance and quantized bits.