Research On Massive MIMO Channel Estimation Based On Compressed Sensing Reconstruction Algorithm

Massive MIMO technology is one of the key technologies of the fifth generation mobile communication system(Fifth Generation,5G).Compared with traditional MIMO multi-antenna systems,massive MIMO multi-antenna systems have many performance potentials.However,the premise of achieving these performance advantages is that the base station already knows the uplink and downlink channel state information.The scale of the number of antennas in a massive MIMO system will increase to hundreds or thousands.Traditional channel estimation techniques are no longer applicable due to the large pilot overhead and low estimation accuracy.Therefore,it is particularly important to study new channel estimation techniques.The research work in this paper is as follows:(1)In a massive MIMO communication system,the channel matrix gene rated when the number of antenna arrays is large will exhibit sparse properties.Massive MIMO systems can use this feature to transform the channel estimati on problem into a reconstruction problem in Compressed Sensing(CS).Firstly,this paper adopts a sparsity adaptive piecewise orthogonal matching pursuit al gorithm(SASt OMP).This algorithm is based on the existing piecewise orthogo nal matching pursuit(St OMP)algorithm and introduces the retrospective idea.A new identification parameter is introduced to achieve effective secondary sup port set screening and sparsity adaptive estimation.Simulation results show th at the channel estimation performance of the algorithm used is significantly bet ter than the OMP reconstruction algorithm and the St OMP reconstruction algori thm.(2)The SL0 reconstruction algorithm is used to transform the channel esti mation problem into an extreme value problem for solving the smooth function.The smooth function is used to approximate the norm and combined with the convex optimization idea.The iterative process uses the steepest descent meth od and the gradient projection principle to gradually approximate the optimal s olution of the channel estimation matrix.Furthermore,this algorithm is further improved.The hyperbolic tangent curve is used instead of the Gaussian functio n for channel estimation,and a modified Newton method optimization algorith m is introduced during the iterative process.Simulation results show that comp ared with commonly used CS reconstruction algorithms,the SL0-based reconstruction algorithm requires less computational complexity and higher channel esti mation accuracy without the need for a known sparseness of the massive MIM O channel.As a result,the channel estimation performance of massive MIMO systems has been significantly improved.