Random Access Signal Detection Algorithm For Massive MIMO Systems

In recent years,with the continuous development of mobile Internet and Internet of Things application requirements,the demand for wireless communication rate and the number of terminal connections have shown an exponential growth.Massive Multi-Input Multi-Output(MIMO)technology has become the core technology of next-generation wireless communication because it can significantly improve the system's power efficiency,spectrum efficiency,rate capacity,and user terminal capacity.This paper studies the codebook scheduling and multi-user detection techniques in massive MIMO random access systems for massive MIMO random access scenarios.Firstly,for the optimal detection / estimation of the most common linear model in signal detection / estimation,a variety of message passing algorithms are studied to solve the edge posterior probability density function in the optimal detection / estimation problem.Starting from the factorization of the joint posterior probability density function,Gaussian approximations are made to each factor of the posterior joint probability density function to derive the expected propagation algorithm.Then on the basis of the expected propagation algorithm,avoid the explicit computation to reduce the complexity of the part with high computational complexity,and approximate the matrix inversion to derive the approximate expected propagation algorithm.In addition,Based on the expected propagation algorithm,for massive MIMO systems,the algorithm is approximated by eliminating some higher-order terms,and an approximate message passing algorithm is derived.Simulation results show that expectation propagation(EP)algorithm and approximation expectation propagation(AEP)algorithm with different initialization methods have similar bit error ratio(BER)performance after algorithm convergence,and The approximate expectation propagation algorithm initialized using the minimum mean squared error(MMSE)method has higher computational complexity in the initialization process,but also has a faster convergence speed.Secondly,the sparse code multiple access(SCMA)codebook design method suitable for massive MIMO systems and the codebook scheduling method when the number of access users far exceeds the number of codewords in the SCMA system are studied.Based on the SCMA codebook design problem model,the design schemes of the expansion matrix and the constellation diagram are studied separately.With the increase of the number of access users,the number of codewords in the SCMA system will not be able to meet the system requirements,and the decoding performance at the receiving end will also be affected.Therefore,in the case where the number of codebooks in the SCMA system is less than the number of system users,the users in the system are grouped and codebook scheduled by analyzing the correlation of beam domain channels between different users.Simulation results show that when the number of access users is less than the number of system codewords,the proposed SCMA codebook design scheme has better performance than traditional SCMA codebooks;while when the number of access users is much larger than the system load,the proposed The codebook scheduling algorithm based on the beam domain channel has better performance than the traditional codebook scheduling scheme.Finally,the uplink multi-user detection method for massive MIMO non-orthogonal multiple access with channel estimation errors is studied,and a robust approximate message passing(RAMP)algorithm for nonorthogonal multiple access(NOMA)in code domains is proposed under the theoretical framework of minimizing Bethe free energy.For a given pilot structure and channel estimation method,the existing signal detection methods directly use the estimated value of the channel as the input of the signal detection algorithm.However,the channel estimation result often has some errors,which will greatly limit the performance of the signal detection algorithm.So using the probability density function instead of the channel estimate as the input of the signal detection algorithm will make the signal detection result more accurate.The problem of uplink multi-user signal detection in the code domain non-orthogonal multiple access system is transformed into a signal detection problem in a multilevel generalized linear model,and based on the probability density function of channel state information(CSI)obtained by channel estimation,Under the Bethe free energy theoretical framework,the problem of large-scale MIMO non-orthogonal multiple access signal detection with channel estimation errors is transformed into a constrained minimized Bethe free energy problem.Aiming at this optimization problem,a series of stagnation point equations are obtained using the Lagrange multiplier method,and a robust approximate message passing algorithm based on non-orthogonal multiple access is derived by solving the stagnation point equations.On this basis,the NOMA-RAMP soft-input soft-output(SISO)detection algorithm that can be used for iterative detection decoding is also studied.Simulation results show that the proposed robust approximate message passing algorithm has better performance and improved robustness to channel estimation errors than the improved generalized approximate message passing signal detection algorithm without increasing the complexity of the algorithm.