Uplink Channel Estimation For Massive MIMO Systems Based On Channel Sparse Representation

With the rapid growing demand for wireless communication services,large-scale multiple input multiple output(MIMO)technology has received more and more attention.Massive MIMO technology is equipped with a large number of antennas on the base station side and serves multiple user terminals at the same time,which can effectively improve system spectral efficiency and data link reliability.The performance of largescale MIMO wireless transmission depends on that of channel estimation.This paper mainly studies the method of large-scale MIMO uplink channel estimation based on channel sparse representation.Firstly,the narrowband and wideband MIMO-OFDM(Orthogonal Frequency Division Multiplexing)channel models and their statistical characteristics are studied separately.Based on the spatial characteristics of massive MIMO channels,a discrete Fourier transform(DFT)matrix is used to transform the spatial frequency domain channel to the beam delay domain channel.Secondly,the problem of channel energy leakage in DFT sampling is considered.The oversampling DFT matrix and MFOCUSS algorithm is used to convert the spatial frequency domain channel into oversampling beam delay domain channel.Then its statistical channel characteristics is studied.The simulation results show that as the oversampling factor increases,the more concentrated the channel beam energy,the more sparse the channel represents.Secondly,the problem of oversampling beam-delay domain channel estimation in narrowband and wideband channels are studied separately.Narrowband channel estimation is solved by using the MFOCUSS algorithm.In the minimum mean square error(MMSE)channel estimation algorithm,the channel estimation problem of all statistical channel information known and only diagonal statistical channel information known are studied.In the sparse Bayesian learning algorithm,based on the Bethe free energy minimization criterion,the channel estimation problem in the case where only diagonal channel information is known is studied.On this basis,further using the sum product algorithm and based on the Bethe free energy minimization criterion,the channel estimation problem under unknown statistical channel information is studied.The simulation results show that as the oversampling factor increases,the channel estimation accuracy increases.In the case of high signal-to-noise ratio,the performance of the MMSE channel estimation algorithm using diagonal statistical channel information is close to that of the MMSE channel estimation algorithm using all statistical channel information.In addition,the channel estimation result of the sparse Bayesian learning algorithm using diagonal statistical channel information is consistent with the MMSE channel estimation result using diagonal statistical channel information.And with the improvement of the signal-to-noise ratio,the performance of the sparse Bayesian learning channel estimation algorithm with the statistical channel information unknown is gradually approaching the performance of the diagonal statistical channel information known.Finally,the effect of the pilot scheduling on the performance of channel estimation based on sparse representation is studied.First,the adjustable phase shift pilot(APSP)scheduling method is studied.On this basis,the channels in the beam domain and the beam delay domain are estimated based on the MMSE criterion respectively.For APSP pilots,different users adjust the pilot phase shift factor dynamically,thereby improving the available pilot resources.Then,on the basis of the single OFDM symbol APSP pilot,the problem of multi-OFDM symbol APSP pilot channel estimation under oversampling is studied.The the maximum adjustable phase shift of multi-OFDM symbol APSP pilot is further extended with respect to the case of single OFDM symbol.The simulation results show that the performance of channel estimation using the APSP pilot scheduling method is close to the performance of channel estimation without pilot interference.The channel estimation performance of APSP pilot scheduling in the case of multiple OFDM symbols is significantly improved compared to the case of a single OFDM symbol.And based on channel oversampling,the performance of APSP pilot channel estimation in multiple OFDM symbols can be further improved.