Compressive Sensing-Based Channel Estimation For MIMO Systems

Multiple-input multiple-output(MIMO)technique can obviously improve the data transmission rate and the spectrum utilization by providing the spatial diversity gain or multiplexing gain,and thus has been widely used in modern broadband wireless communication systems.To achieve the theoretical performances gains,the accurate channel estimation at the receiver is becoming essential for MIMO systems.Based on the compressive sensing theory,the hidden sparsity of the wireless communication system is fully utilized in this thesis and several improved channel estimation algorithms are also proposed for various MIMO system models.The specific contents are structured as follows:1.Due to the problems that that the existing researches of MIMO-OFDM channel estimation do not fully consider the factors especially for the channel correlation characteristics that may have an effect on system performance,a channel model utilizing the spatial and temporal sparsity is established and a spatial-temporal channel estimation algorithm is also proposed in this thesis.Firstly,the proposed algorithm estimates the sparse support set according to the principle of maximum correlation,and then the iterative estimation of channel impulse response is processed by using the LMMSE algorithm.Simulation results show that the spatial-temporal channel estimation algorithm can effectively improve the estimation accuracy and recovery probability,and the relationship between the number of receiving antennas and time correlation parameters is also ilustrated.2.Considering the multi-user MIMO-OFDM systems based on the proposed space-time model,as the existing researchs do not consider the existence of sparse channel correlation among users,a joint sparse estimation algorithm is proposed in this thesis.Firstly,the proposed algorithm estimates the common support set for all users according to the principle of maximum correlation,followed by the processing of spatial-temporal channel estimation algorithm to estimate the individual support set and the channel impulse response of different users.Simulation results show that the joint sparse estimation algorithm can make full use of the hidden sparsity of channel and thus improves the channel estimation accuracy,and the influence of joint sparsity on estimation performance is also verified.3.Aiming at the problem that the pilot overhead of channel estimation becomes overwhelming for massive MIMO systems,a joint estimation scheme combining downlink channel estimation and uplink feedback is applied in this scenario.By utilizing the rank sparsity of channel matrix,the joint channel estimation problem can be formulated as a low rank matrix completion problem.Then a soft thresholding method is applied based on the model by selecting an appropriate threshold to achieve the ideal performance of channel estimation.Simulation results show that,compared to the existing algorithms of low rank matrix estimation,the proposed algorithm reduces the pilot overhead and achieves higher estimation accuracy and faster convergence meanwhile.