Research On Principles And Key Technologies Of SCMA

More terminals,higher spectrum efficiency,lower latency,and ultra-high reliability indicators pose huge challenges to the fifth generation(5G)and future mobile communication systems.The new non-orthogonal multiple access(NOMA)schemes represented by superimposed transmission no longer require orthogonality,and more user information can be superimposed on the same resource elements.Compared with the traditional orthogonal schemes,NOMA schemes have obvious technical advantages in terms of enhanced spectrum efficiency,improved user connection capabilities,and reduced air interface transmission delay,which can be applied to different typical 5G application scenarios.Among the many NOMA schemes,sparse code multiple access(SCMA)is a very promising NOMA scheme.However,its codebook optimization problem is still a complex multi-dimensional problem where the optimal solution cannot be obtained.In addition,in scenarios with large codebook size,multiple antennas,or high-density,the detection complexity at the receiver is still very high.In order to solve the shortcomings of the current SCMA scheme,this dissertation takes the principles and key technologies of SCMA scheme as the core,and focuses on how to improve the reliability of the scheme and reduce its detection complexity.Specifically,in order to improve system reliability,SCMA codebook and factor graph matrix are optimized.To reduce the detection complexity,a linear complexity detection algorithm is proposed.The research content of this dissertation fully taps the potential of SCMA technology,which not only improves the reliability,but also reduces the detection complexity,which helps to promote the application of this technology in 5G and future communications.This dissertation focuses on the following aspects.First of all,in order to have a deeper understanding of the SCMA scheme,this dissertation systematically introduces the SCMA principles and the related key technologies.This dissertation respectively introduces and compares the uplink and downlink SCMA system models.Then the key technologies such as SCMA codebook design,factor graph matrix characteristics and detection algorithm are introduced.In terms of the codebook design,this dissertation introduces a step-by-step optimized suboptimal SCMA codebook design method.In tenns of the detection algorithm,this dissertation first introduces the maximum a posteriori(MAP)algorithm with the best performance but higher complexity.Subsequently,the log-domain message passing algorithm(Log-MPA)and Max-Log-MPA,which can significantly reduce the detection complexity are introduced.Through simulation results,this dissertation verified that SCMA scheme has significant gains compared with the low-density signature(LDS)scheme and the orthogonal frequency division multiple access(OFDMA)scheme,and the gains increase with the increase of the overloading rate or the codebook sizes.Secondly,this dissertation proposes two SCMA codebook optimization methods for the uplink and the downlink SCMA system models.Considering the current optimization problem on the multidimensional codebook has not yet obtained an optimal solution,so this work helps to provide guidance for SCMA codebook optimization.For the uplink system,this dissertation proposes an uplink SCMA codebook optimization algorithm which combines geometric shaping and probabilistic shaping.The algorithm takes the maximum average mutual information(AMI)as the optimization criterion,and utilizes bare bones particle swarnm optimization(BBPSO)algorithm to jointly optimize the SCMA codebook with non-uniform distribution.In order to avoid the BBPSO algorithm falling into the local optimal solution,this dissertation adjusts the topology of the particle swarm.Considering that there is no closed-form expression for the AMI,and the Monte Carlo simulation method has high complexity,this dissertation derives the closed form expression of the lower bound of AMI.Based on the downlink SCMA system,this dissertation proposes an SCMA codebook optimization algorithm with lower complexity,which takes the lower bound of the AMI as the optimization criteria.Unlike the existing literature,which optimize the mother codebook and the rotation angles between different codebooks separately,the algorithm jointly optimizes the mother codebook and the rotation angles.For different channels,overloading rate and codebook sizes,this dissertation proves the superior performance and robustness of optimized codebooks from theoretical analysis and computer simulations.And it can be obtained through simulation results that with the increase of overloading rate or codebook sizes,the gains of the optimized codebooks compared with the reference codebooks are more obvious.Then,based on the traditional expectation propagation algorithm(EPA),this dissertation proposes the Max-Log-EPA multi-user detection algorithm with linear complexity for SCMA system.Considering that the high detection complexity of the receiver is a major obstacle to the practical application of SCMA technology,especially for scenarios with large codebook size,multiple antennas,or user-intensive scenarios,this work helps to improve the engineering practicality of SCMA technology.The Max-Log-EPA algorithm not only reduces the complexity by using the logarithmic operation,but also utilizes the Max operation to replace part of the logarithmic operation and exponential operation,which can further reduce the complexity.This dissertation compares and analyzes the complexity performance of the proposed algorithm with the typical detection algorithms of SCMA scheme.The analysis results show that with the increase of the number of receiving antennas,superimposed users or codebook sizes,the greater the reduction in the computational complexity of our proposed algorithm.In addition,this dissertation compares the block error rate(BLER)performance of the Max-Log-EPA detection algorithm with the typical SCMA detection algorithm under different parameter settings.The simulation results prove that our proposed algorithm has a slight performance loss at low signal-to-noise ratio(SNR)compared to the traditional Max-Log-MPA detection algorithm,which is about 0.1dB?0.2dB,and almost no performance loss at high SNR.Finally,this dissertation optimizes the factor graph matrix of SCMA and constructs a large girth and irregular large-scale factor graph matrix.Considering that both the SCMA codebook optimization and detection algorithm are based on factor graph matrix,the optimization of factor graph matrix is also the focus of this dissertation.The sparse factor graph matrix of SCMA and the check matrix of low-density parity check(LDPC)code have extremely similar characteristics,and the idea of SCMA's MPA detection algorithm is also derived from the belief propagation(BP)algorithm of LDPC code.Therefore,this dissertation draws on the idea of optimizing the LDPC code check matrix and studies the design of the SCMA factor graph matrix.A large-scale matrix can disperse user's data in channels,thereby increasing diversity gain.Irregular degree distribution can meet the different needs of different users on the number of occupied resources,and bring coding gain.Large girth can improve the performance of iterative detection algorithm.In this dissertation,the unequal protection criterion is used to optimize the degree distribution of the factor graph matrix in the near-far effect scenario,and the progressive edge growth(PEG)algorithm is used to construct a large girth and irregular large-scale SCMA factor graph matrix.The advantages of the constructed matrix over the reference matrix in iterative detection are analyzed by extrinsic information transfer(EXIT)chart.Based on this factor graph matrix,a low-density superposition modulation(LDSM)scheme that can flexibly set the number of SCMA data layers occupied by a single user is proposed.Both BLER simulation results-and EXIT analysis prove that the proposed LDSM scheme has a significant performance gain over the reference scheme.In short,this dissertation systematically introduces the principles and key technologies of the SCMA scheme and proposes corresponding optimization algorithms,which fully taps the potential of this NOMA scheme.It helps to promote its engineering applications,and can further meet the needs of 5G and future communications such as large connections,low latency,and high spectral efficiency.