| In millimeter wave(mm Wave)massive multiple input multiple output(MIMO)systems,there is a certain error between the phase of the carrier signal and the phase of the local oscillator,which is called phase noise.The existence of phase noise will seriously affect the performance of systems,especially in orthogonal frequency division multiplexing(OFDM)systems,which will destroy the orthogonality between subcarriers,which is the main cause of the degradation of the performance of OFDM wireless communication systems.At present,many channel estimation schemes ignore the existence of phase noise,but in wireless communication systems,phase noise is real,and the damage of phase noise is usually unpredictable and uncontrollable.In this study,the problem of joint estimation of channel and phase noise in mm Wave massive MIMO-OFDM system is mainly solved under the influence of common phase noise and independent phase noise,and the effective joint estimation schemes of phase noise and channel are proposed to reduce the signal distortion caused by phase noise.The main research contents of this paper are as follows:(1)A joint phase noise and channel estimation scheme based on pattern search and compressed sensing is proposed.Firstly,a mm Wave massive MIMO-OFDM system model with common phase noise is established,and a time-varying training mode is proposed,which can save a lot of training time compared with the time-invariant training mode.Then for the case where both the phase noise and the channel are unknown,a solution of phase noise and channel estimation is proposed,which uses the pattern search method to estimate phase noise and the orthogonal matching pursuit method to estimate the channel.The estimated result will converge according to the residual of the received signal,so as to optimize the performance of channel estimation.(2)A joint estimation scheme of phase noise and channel based on deep learning is proposed.Firstly,a mm Wave massive MIMO-OFDM system model with independent phase noise is established.Then,a feedforward neural network(FNN)was designed and trained to estimate phase noise in time domain.The designed FNN can estimate phase noise in time domain online after off-line training.The obtained time domain phase noise estimation results can be used in the final channel estimation. |
