Research On Low-complexity Multi-user Detection Algorithm In SCMA System

To meet the diversified performance requirements of mobile internet and internet of things,and to reach the vision of all things connected towards 2020 and beyond,the 5th generation mobile communication network(5G)need to have the characters of massive connection and low delay access.Sparse code multiple access(SCMA)technology can make the system maintain good performance under high overload rate,and now,it has become a very competitive candidate multiple access scheme in 5G.Different from the traditional modulation and spreading spectrum processing,the coded bits are directly mapped into complex domain multi-dimensional sparse codewords at transmitter side in SCMA,and the codewords of different users are superposed on the same physical resource blocks in a non-orthogonal way through sparse spreading at receiver side.The message passing algorithm(MPA)can be used for multi-user joint detection based on the sparse character of codewords.This thesis conducts thorough research and analysis on SCMA in terms of codebook design and multi-user detection,then some improved schemes are proposed,the main contribution of this thesis can be concluded as follows:1.This thesis firstly analyzes the evolution route of SCMA system,and points out the feasibility of achieving massive connection in the scheme.Then the basic principle of uplink SCMA system is analyzed in detail,including both the transmitter side and the receiver side.Finally,the factors that affect performance of system are analyzed,including the number of iterations and the system overload rate.Research results show that compared with long term evolution(LTE),SCMA can achieve 300% overload rate and provide better connectivity.2.Aimed at the complicated design problem of SCMA codebook,the steps of codebook design are firstly studied.Then a novel codebook design method based on Star-QAM constellation is introduced in this thesis,which can make the power of each codeword and dimension different in the codebook.Furthermore,it can maximize the average Euclidean distance,thereby the system bit error rate(BER)performance can be enhanced.Research results show that the performance of codebook based on Star-QAM constellation is better than that of proposed by Huawei 5G contest.3.Aimed at the high decoding computation and poor performance problem of original MPA under high overload conditions in SCMA,two novel improved algorithmsare proposed in this thesis: partial information based on serial of MPA(PIS-MPA)and weighted factor based on sphere decoder of MPA(WFSD-MPA).For the PIS-MPA algorithm,the message is updated in a serial manner,and only some of the external information is updated by setting threshold value.And for WFSD-MPA algorithm,it can realize a better performance by reducing the search range of constellation points and introducing weighted factors to change the initial probability of constellation points in the search range during decoding.Research results show that both the two improved algorithms not only can reduce the decoding complexity,but also accelerate the convergence of iterative process.In addition,they can dynamically achieve the trade-off between computational complexity and performance with suitable threshold parameters choice,which can flexibility meet the performance requirements of different application scenarios in future 5G mobile communications.