Research On MIMO Signal Detection Algorithms For Massive Connections

With the rapid upgrade of mobile communication technology,a large number of smart devices are connected to the Internet.“Massive Connection Era” has brought higher technical requirements to wireless communication.As one of the key technologies of 5th generation mobile communication,massive multiple input multiple output(MIMO)technology has attracted much attention.However,massive connections bring many difficulties to MIMO signal detection.For example,traditional detection schemes cannot balance the relationship between performance and complexity.Therefore,the research focus has gradually shifted to more excellent MIMO signal detectors.In view of the shortcomings of traditional detection technologies,this thesis will fully exploit the system characteristics in narrowband and wideband massive connection scenarios,and conduct in-depth research on high-performance and low-complexity MIMO signal detection algorithms.Firstly,an overview of MIMO signal detection technology is introduced.This thesis first introduces the MIMO system model,and compares the characteristics of the complex model and the equivalent real model.Then,focusing on the traditional linear and nonlinear detection algorithms,this thesis clarifies the basic principles of MIMO signal detection,and compares the performance and complexity of different algorithms.Finally,the contradiction between the performance and efficiency of traditional signal detectors is discussed,which helps to clarify the research direction of this thesis.Additionally,this thesis studies sparse signal detection algorithms in the narrowband massive connection scenario.The grant-free random access technology in narrowband massive connection scenario is studied,and the grant-free random access system is modeld in this thesis.Then,the thesis proves the equivalence of the active users detection problem and the compressed sensing problem.Combined with compressed sensing theory,the convex optimization,greedy and Bayesian sparse signal detection algorithms for multiple input single output system and multiple input multiple output system are presented respectively.Finally,this thesis verifies the performance of active users detection and channel estimation for different sparse signal detection algorithms through simulations.Furthermore,the sparse aware detection algorithms for wideband massive connection MIMO system are studied.This thesis first discusses two methods of constructing sparseness for non-sparse signal,namely the method of high-dimensional sparse decomposition and sparse error vector,and then compares the two methods with each other in terms of their advantages and disadvantages.Secondly,based on the method of sparse error vector,a sparse error maximum a posteriori probability detection algorithm is proposed.By constructing an error recovery optimization problem and transforming it into a convex quadratic minimization problem,a closed-form solution is obtained and an iterative detection scheme is derived.Then,this thesis uses the method of orthogonal decomposition for error vector to generalize the proposed algorithm to high-order modulation scenarios,where the generalized sparse error maximum a posteriori probability detection algorithm is acquired.Finally,by analyzing the results of simulations,this thesis fully verifies that the performance of the proposed algorithm is significantly better than the traditional algorithms.Finally,this thesis studies the detection algorithms based on approximate message passing for wideband massive connection MIMO system.This thesis first points out that the approximate message passing algorithm can directly exploit the structure of the signal itself.Then,the similarities and differences of two traditional approximate message passing algorithms are analyzed,which explain the limitations of their application in wideband massive connection MIMO system.Secondly,an approximate message passing algorithm for structured signal is proposed,which transforms the MIMO signal detection problem into a basis pursuit de-noising problem.Using the central limit theorem and Taylor expansion to simplify the belief propagation algorithm,this thesis gives the approximate iterative rules within the limits of the large system.Subsequently,an iterative damping strategy for convergence is studied,which lowers the error floor of the proposed algorithm.Finally,the significant superiority of the proposed algorithm in performance and complexity is verified by simulations.