Application Of Compressed Sensing Technology With Temporal Correlation In FDD Massive MIMO

Massive multiple-input multiple-output(MIMO)systems which employ a large number of antennas at the base stations can serve many users with the same time-frequency resource.It can significantly improve the spectrum efficiency,channel capacity and link reliability of the wireless communication systems and meet the needs of future wireless communication services.In order to fully utilize the advantages of massive MIMO,the base station has to obtain accurate downlink channel state information(CSIT)for operations such as beamforming,precoding,resource allocation and so on.In frequency-division duplexing(FDD)systems,channel state information(CSI)can be achieved through downlink pilot training and uplink feedback,the overhead used for downlink channel estimation and feedback usually scales proportionally to the number of antennas at the base station,which consumes large amounts of communication resources.This thesis focuses on the channel sparsity and temporal correlation of FDD massive MIMO systems,researches the compressed sensing technology and applies it to channel estimation and feedback.Firstly,the key technologies,channel estimation methods and compressed sensing theory in mobile communication systems are summarized.The principles,advantages and challenges of MIMO,massive MIMO and orthogonal frequency division(OFDM)are briefly described.Then,the downlink channel estimation methods including least square(LS)and minimized mean square error(MMSE)are introduced.After that,the compressed sensing theory is described detailedly in three parts: the sparse representation of signal,the design of measurement matrix and signal reconstruction algorithms.Moreover,two channel estimation methods and common signal reconstruction algorithms are simulated and analyzed.Secondly,the compressed sensing channel estimation scheme is investigated based on the temporal correlation of channels.Based on the sparsity and temporal correlation of channels,the system model and the universal channel sparse model are provided.Then,we discuss a channel estimation algorithm called modified subspace pursuit(MSP)algorithm,which can exploit the prior channel support.Aiming at the scenario of model mismatch,we discuss the conservative MSP algorithm,and considering its performance degradation in the scenarios without model mismatch,we propose an adaptive MSP algorithm.After that,the above algorithms are simulated,analyzed and compared.The simulation results show that the MSP algorithm has performance gain over conventional signal reconstruction algorithms.The conservative MSP algorithm and the adaptive MSP algorithm are more robust to model mismatch scenario.The adaptive MSP algorithm has better adaptability to the channel propagation environment.Finally,the adaptive channel estimation and feedback scheme are investigated based on the common sparsity of channels in time-frequency domain.In massive MIMO-OFDM systems,the common sparsity in frequency domain and the temporal correlation of channels are analyzed.Additionally,the channel estimation and feedback scheme is summarized.Then,based on the common sparsity of channels in frequency domain,we discuss the distributed sparsity adaptive matching pursuit(DSAMP)algorithm which can estimate the channels of multiple pilot subcarriers jointly,and propose the design of non-orthogonal pilots.Based on the DSAMP algorithm,we discuss the adaptive CSI acquisition algorithm which can adaptively adjust the pilot time slot length to obtain reliable CSI.After that,based on the temporal correlation of channels,we discuss the closed-loop channel tracking scheme which can estimate the channels in consecutive multiple time blocks with less pilot slot overhead.The above algorithms are simulated,analyzed and compared.The simulation results show that the DSAMP algorithm has better channel estimation performance than the conventional signal reconstruction algorithms without sparsity level as prior information.The adaptive CSI acquisition algorithm can adaptively adjust the pilot time slot length to an appropriate value,and its performance is close to the upper bound.The closed-loop channel tracking scheme further reduces pilot time slot overhead while maintaining good channel estimation performance.