Research On Massive MIMO Channel Estimation Algorithms Based On Structured Compressive Sensing

In order to cope with the demand of higher data rate and lower delay in the fifth generation mobile communication system(5G),massive Multiple-Input Multiple-Output(MIMO)technology has been put forward and studied widely.Massive MIMO technology has the advantages of high spectrum efficiency.It can improve the capacity of multiuser network by a large margin.However,with the increase of the number of antennas on the base station of massive MIMO,the complexity of its channel estimation will increase.The traditional channel estimation algorithms,such as least squares algorithm and least mean square algorithm,are no longer applicable to massive MIMO systems.It is another effective way to solve its channel estimation problem by applying compressed sensing to massive MIMO system for channel estimation.It can also achieve good channel estimation performance while reducing pilot overhead and algorithm complexity.There are many shortcomings of existing sparse channel estimation algorithms for massive MIMO systems based on compressed sensing,such as long computation time and high computational complexity,we studied the sparse channel estimation algorithm for massive MIMO systems.(1)The characteristics of the large scale fading and the small scale fading of the wireless channel are studied.Analysis model characteristics and channel capacity of MIMO system,and analysis system model,channel capacity and convergence of massive MIMO system,and through the simulation,the massive MIMO channel matrix gradually orthogonality theory is verified by simulation.(2)Firstly,including the Least Squares(LS),Minimum Mean Square Error(MMSE)algorithm,the channel estimation algorithms of traditional MIMO system are studied.The principle of two algorithms is described in detail to illustrate the non applicability of the two algorithms in massive MIMO systems.Secondly,the author summarizes the research progress of compressed sensing theory,and uses the sparse characteristic in the angle domain to deduce the formula of downlink and summarize the system sparse channel estimation algorithm flow of massive MIMO system.Then,the author analyses the step of channel estimation algorithm of the classical Orthogonal Matching Pursuit(OMP).Because the sparsity of the wireless channel is not easy to be accurate,the accuracy of the OMP algorithm can be degraded.The author puts forward the improved Sparsity Adaptive Matching Pursuit(SAMP)channel estimation algorithm.Finally,the performance analysis and simulation verification of the algorithm are carried out.Compared with the traditional channel estimation algorithm,theoretical analysis and simulation results show that the sparse channel estimation algorithm can greatly reduce the pilot cost and improve the accuracy of channel estimation.It has more practical value.(3)Based on the above research results,the author studies the theory of structured compressed sensing.When the wireless channel is structured sparse,it can get better estimation performance by using the structured compressed sensing theory to estimate the channel.First of all,in the pilot distribution scheme for traditional channel estimation,it has the phenomenon of large pilot cose and low spectrum utilization in massive MIMO system.A new massive MIMO channel estimation pilot distribution scheme(non orthogonal pilot)is presented.Then,by the analysis of the space-time correlation characteristics of the time-delay domain of massive MIMO systems,it can advances a massive MIMO structured sparse channel estimation algorithm that based on space-time correlation.The problem of overestimation is caused by the slow growth of each iteration step in the algorithm.In this algorithm,the Variable step size Adaptive Structured Subspace Pursuit(VssASSP)algorithm is proposed by introducing the thought of variable step length and setting two iterative stopping criteria.This operation can guarantee the accuracy of channel estimation.Through algorithm analysis and simulation verification,it is shown that under the same conditions,the proposed algorithm can effectively reduce computational complexity and computation time compared with its similar algorithm.In summary,the theoretical analysis and simulation results show that the proposed algorithm has great theoretical value and potential application prospects in the future 5G applications.