Time-varying Channel Estimation For Massive MIMO-OFDM System

Massive MIMO-OFDM system is a combination of massive multiple-input multipleoutput(MIMO)system and orthogonal frequency division multiplexing(OFDM)modulation technology.Massive MIMO technology can greatly increase the throughput of the system by greatly increasing the number of antennas at the transceiver end.The combination with OFDM technology can improve the ability of massive MIMO system to resist multipath interference and simplify receiver design.Due to its significant advantages of high rate,low latency,and high reliability,massive MIMO-OFDM technology has become one of the key technologies of the next generation wireless communication system.In actual wireless communication scenarios,the movement of either the transmitter,receiver,or reflector will cause the Doppler effect,shift the frequency of the transmitted electromagnetic waves,and introduce time-varying characteristics for the channel.In a system based on OFDM modulation,the Doppler effect will destroy the orthogonality between sub-carriers,resulting in inter-carrier interference(ICI),causing bit errors at the receiving end.In a massive MIMO-OFDM system,the antenna array is large in scale,and the working frequency band of each antenna is the same,and the ICI caused by the Doppler effect will increase.In order to eliminate the ICI introduced by the Doppler effect,the receiving end needs to cancel the role of the channel through equalization.Therefore,in a time-varying scenario,how to efficiently and stably obtain channel state information(CSI)has profound theoretical research significance and practical application value.This paper studies the channel estimation of massive MIMO-OFDM systems in timevarying scenarios.In this work,at the level of OFDM data symbols,a complete model of a massive MIMO-OFDM system under a time-varying channel is given,and the problem model of channel estimation is constructed accordingly.Based on the joint block sparsity of the time-varying MIMO channel in the time-delay-Doppler-angle three-dimensional domain,we attribute this channel estimation problem to the problem of compressed sensing for block-sparse signals.In response to this problem,we propose a new compressed sensing algorithm-Variance State Propagation(VSP).VSP is a Bayesian learning-based compressed sensing algorithm.The layered Gaussian probability model combined with Markov Random Field(MRF)can accurately describe the block sparsity in the signal to be estimated.VSP also uses message passing technology to make the algorithm more efficient.In view of the model mismatch problem that may be caused by the discretized channel model,we use the Expectation Maximization(EM)algorithm to learn the discretized grid parameters,modify the measurement matrix in the compressed sensing problem,and greatly improve the presence of VSP in the channel Performance under model deviation scenarios.The VSP algorithm can effectively reduce the pilot load required when estimating a time-varying massive MIMO-OFDM channel,and provides excellent estimation accuracy.Simulation results show that under different pilot symbol numbers and noise levels,the VSP's channel estimation performance exceeds that of other comparison algorithms.