Research On The Sparse Code Multiple Access Technology For 5G

Sparse code multiple access(SCMA)is one of the novel non-orthogonal multiple access technologeies facing the fifth generation(5G).Compared with traditional orthogonal multiple access,SCMA performs a higher spectral efficiency and surports more network connections,which makes it the research hotspot in the future airinterface technologies.Differing from the low density signature(LDS),SCMA modulates binary data into complex multi-dimensional codewords directly at the tassmit end,combining mapping and spreading together.The shaping gain producted by the multi-demensional constellation improves the performance of SCMA in comparison with LDS.For the sparsity of codewords,the reciever takes advantage of message passing algorithm(MPA)to multi-user detection(MUD)and shows a excellent performance even when the system is heavily overloaded.MPA is an iterative algorithm based on the factor gragh,in which the extrinsic imformantions are passed between the variable nodes(VNs)and function nodes(FNs)successively.Finally,a stable probability distribution is abtained and used as a detection criterion to choose the codeword with maximum value as the output.MPA is a suboptimum method for the joint optimum maximum-a-posterior(MAP)algorithm,however,benenifiting from the sparsity of codebook,MPA reduce the complexity of MUD tremendously.Although the complexity of MPA is lower than MAP,it is unberable for hardware implementation when the system is heavily overloaded or the size of codebooks is too large.There are many schemes to reduce the complexity of MPA,sush as weighted message passing algorithm(WMPA),shuffled message passing algorithm(SMPA),SCMA codebook design with low number of projections and so on.All these schemes can be devided into two categories,the improvement of MPA and the optimization of codebook.This paper focuses on the improvement of MPA and proposes several schemes in both uplink and downlink system.The main work of this paper is organizes as follows:(1)A low complexity MPA based on threshold is designed in the uplink SCMA system.In the modified MPA,a threshold is applied to choose the codewords with high reliability to decode after each iteration.Thanks to the success interference cancellation(SIC)property of codebook,the users with high energy in the colliding users can be detected soon and they don't need to iterate untill the maximun number of iteration is reached.Compared with the original MPA,the proposed method is complexity-saved for its smaller average number of iteration of users when the signal-to-noise ratio(SNR)is relatively high.Meanwhile,the proposed scheme shows a similar performance with original MPA when the threshold is set appropriately.(2)On the basis of SMPA,a low complexity SMPA based on compressed codebook is put forward,in which the fast convergence of SMPA helps to abandon the codewords with low reliability in time and reduce the search range of the message update in the next iterations.(3)In the downlink SCMA system,we propose a modified MPA,which just aims at part of the recieved signal including the resources ralated with the user and only passes message along a small number of edges in the factor gragh.Consequecly,the complexity of each iteration is reduced compared with the otiginal MPA.Moreover,the performance of the new MPA is better than the original MPA without iteration for the probability distribution of codewords after several iteration is superior to the initialization.