Research On Sparse Code Multiple Access Technology Based On Multi-stage Optimization Strategy

In order to meet the massive connection demand of 5 Q the Sparse Code Multiple Access(SCMA)technology,which evolves from Low Density Signature(LDS)technology,is becoming a hot research topic.By merging the modulator and sparse spreader in the LDS,the SCMA system relies on the codewords in the pre-allocated codebooks to distinguish the data of each user.Benefiting from the sparseness of the codewords,the SCMA system can decode with MPA that takes into account both detection complexity and detection performance.For the key technologies of SCMA system,this thesis studies in depth the codebook design scheme based on multi-stage optimization strategy and the low complexity optimization method of Message Passing Algorithm(MPA).In the codebook design based on the multi-stage optimization strategy,the complexity of designing the mother constellation and the difficulty of determining the optimal constellation operators actually increase the difficulty of the codebook design.To solve this problem,this thesis proposes two improved codebook design schemes,namely the codebook design based on resource block constellation and the codebook design based on distance spectrum.1.Codebook design based on the resource block constellation.By directly designing the low-dimensional resource block constellation with the largest minimum Euclidean distance,the proposed scheme avoids the complexity of designing the high-dimensional mother constellation,and the maximizing minimum Euclidean distance between constellation points on the resource block also improves system performance.2.Codebook design based on distance spectrum.The proposed scheme transforms the design of the mother constellation and the determination process of the constellation operators into a simple codebook design parameter selection process,and reduces the complexity of parameter selection in a way that does not affect the unique decodability of codebooks.Under the codebook design criterion of maximizing the minimum element of the codebook distance spectrum,the proposed scheme determines the optimal codebook design parameters,and finally obtains the user codebooks with excellent performance.The simulation results show that the bit error rate(BER)performance of the two codebook designs proposed in this thesis is significantly improved,compared with the codebook design based on the multi-stage optimization strategy.Aiming at the problem of poor convergence performance of the original MPA,this thesis focuses on the serial structure MPA with good convergence performance and the most practical hybrid structure MPA.According to the fact that the message from the resource node to the user node gradually converges with the increasing of number of iterations,this thesis proposes a low complexity optimization method based on update rate.The proposed method judges whether to stop updating the converged message early by the update rate of the resource node to the user node message,thereby reducing the detection complexity of the MPA.Combined with this method,this thesis proposes a serial structure MPA based on update rate and a hybrid structure MPA based on update rate.The simulation results show that the proposed algorithms can achieve a good balance between detection complexity and BER performance.Furthermore,due to the fact that the grouping strategy plays an important role in the performance of the hybrid structure MPA,this thesis also studies the performance of the hybrid structure MPA under different grouping strategies,and summarizes the optimal grouping strategy under different number of iterations.