Research On Codebook Design And Detection Techniques Of SCMA Systems

The maturing services of internet of things and the emergence of numerous smart devices have brought great challenges to the future mobile communication systems.Restricted by the limited frequency resources,the traditional Orthogonal Multiple Access(OMA)techniques can not handle situations in which huge numbers of users access to the network simultaneously.Therefore,the non-orthogonal multiple access(NOMA)techniques have drawn wide attention and extensive research.Sparse Code Multiple Access(SCMA),as one kind of NOMA,improves the capacity of connectivity significantly by code-domain access,and provides greater diversity gain and coding gain,and the low-complexity detector of SCMA maintains near-optimal performance.Therefore,SCMA technique has attracted much attention and is worth studying.Based on the analysis of basic principles of SCMA,the dissertation focuses on the research of SCMA codebook design and detection techniques.The SCMA codebook design in uplink fading channels,and the low complexity detection algorithms in downlink channels as well as in uplink SCMA systems with multiple antennas have been researched in the dissertation.Firstly,the fundamental of SCMA is introduced.The OMA techniques applied in the first four generations are introduced and their drawbacks are also shown.Then the Low Density Signature(LDS)technique,a code-domain NOMA,and SCMA,derived from LDS are introduced.Then,the SCMA transceiver systems in both uplink and downlink are presented.Next the performance gains of SCMA compared to the OMA and LDS,namely diversity gain and coding gain,are introduced.In addition,the Maximum A Posteriori(MAP)detector which has the best performance,and the Message Passing Algorithm(MPA)detector of which performance is close to MAP detector with a lower complexity,are introduced.Besides,the Log-MPA which performs MPA in logarithm domain is presented.The performance and complexity of the above three detectors are analyzed and compared.Then,SCMA codebook design algorithms in uplink are studied.On one hand,for cases in which receivers perfectly know the channel conditions of all users,the SCMA performance is analyzed,in which the pairwise error probability(PEP)is derived.Based on PEP,the mother con-stellation construction is modeled as a non-convex optimization problem.Further,the problem is simplified to a second order cone programming(SOCP)problem which can be solved efficiently by optimization tools.On the other hand,for cases in which receivers can not know the chan-nel conditions,a transceiver based on LDS is proposed in this dissertation,in which a differential modulation and demodulation are applied.To reduce the detection complexity,the transceiver is optimized,after which the complexity will not increase as the number of receive antennas in-creases.The performance is also analyzed and simulated which shows that the performance will improve dramatically as the number of receive antennas increases,though the transceiver performs not well as the LDS system with coherent detection.Next,the low complexity detection techniques in downlink is studied.A region-restricted MPA algorithm is proposed based on the idea of sphere decoding,taking advantage of the down-link characteristics.The RR-MPA can achieve near-optimal performance while the complexity is reduced significantly.In addition,the Max-Log-MPA is studied and the reason for its performance loss is also found.Thus an improved Max-Log-MPA is proposed in which the maximum correc-tion item is also considered,and improves the performance while the complexity maintains almost the same.The RR-MPA and the improved Max-Log-MPA are combined together to further make a better compromise between the complexity and the performance.Finally,the low complexity detection techniques in uplink SCMA systems with multiple an-tennas is studied.According to the number of antennas,the system are divided into two cases:overdetermined and underdetermined.For overdetermined SCMA,MMSE-MPA and JGA-MPA are introduced to reduce the complexity with a performance loss;Then a Resource-Selection based(RSB)MPA is proposed in which the minimum singular value of the equivalent channel is used to assess the channel condition,and the RSB-MPA maintains the performance with the complex-ity reduced.For underdetermined SCMA,the Edge Selection(ES)MPA is introduced.Then an Adaptive Edge Selection(AES)is proposed in which the number of selected edges can be ad-justed according to the channel condition,and the compleixty is reduced further.The complexity comparison and simulation are given subsequently.