Compressive Sensing Based Channel Estimation In Massive MIMO

MIMO has become the core of wireless communication,which can significantly improve the system capacity without additional consumption of spectrum resources.In order to obtain greater system gain,massive MIMO was proposed in 2010,which has now become one of the core technologies of 5G.However,the advantage of large scale MIMO is based on accurate channel state information and the huge pilot overhead becomes a difficult problem since the antenna size increases greatly.Compressive sensing,can be used to reconstruct the signal with less data below Nyquist sampling rate,which is widely used in various fields.In order to reduce the number of pilots,compressive sensing is applied in channel estimation.In this thesis,based on compressive sensing,the research on large-scale MIMO FDD channel estimation is organized as follows:Firstly,basic compressive sensing framework is described in detail,including sparse transformation,how to assess measurement matrix and the theory of reconstruction algorithms.Also,some simulation results of the algorithms have been demonstrated.Then,sparse channel impulse response model in time-domain is introduced.Compressive sensing algorithms can exploit the sparse property to reduce pilot length while traditional algorithms can not.Besides,since the antenna array is closely arranged,based on temporal correlation,this thesis proposes block Iterative Hard Thresholding algorithom according to Iterative Hard Thresholding.Different from traditional iterative thresholding algorithms,the proposed algorithm update the iteration through block,which has better accuracy under sanme condition.At last,this thesis describes the angular-domain channel model and compressive sensing is applied to exploit the sparsity of angular-domain channel matrix.In addition,based on the temporal correlation under flat block-fading situation,this thesis proposes modified Sparsity Adaptive Matching Pursuit and modified Iterative hard thresholding,exploiting prior support to narrow element selection range.Compared with original algorithms respectively under same condition,simulation results show that two modified algorithms can be used to improve channel estimation accuracy and reduce pilot overhead by exploiting prior information adaptively.