Research And Design Of Massive MIMO Channel Feedback Based On Compressed Sensing

Massive MIMO technology is one of the core technologies of 5G.The dramatic increase in the number of antennas in massive MIMO systems makes traditional channel feedback techniques unsuitable,and new effective channel feedback schemes must be explored.The theory of compressed sensing is rapidly evolving and widely used in many fields such as image and signal processing.Compressed sensing technology provides a new way for channel feedback of massive MIMO systems with good compression and reconstruction performance of sparse signals.This thesis mainly explores massive MIMO channel feedback based on compressed sensing.The main contributions of this thesis are as follows:Firstly,this thesis studies the channel feedback scheme based on compressed sensing and simulates its performance.The existing channel feedback reconstruction algorithm is researched and analyzed.An improved algorithm,namely GOMP algorithm,is proposed for the existing deficiencies of the reconstruction algorithm,and the channel reconstruction performance of the two reconstruction algorithms is theoretically analyzed and simulated.Secondly,considering that the signals usually have strong correlation in the actual application scenarios,the channel feedback scheme based on distributed compressed sensing is studied.Two application scenarios of single-user and multi-user channel feedback are detailed.For the single-user channel feedback,a channel feedback scheme using space-time correlation of channels is proposed,which improves the channel reconstruction accuracy without increasing the feedback amount.Aiming at the channel feedback under multi-user,a joint sparse model is proposed based on the spatial correlation of the channel and the channel correlation between adjacent users.Based on the model,a new reconstruction algorithm,namely distributed generalized orthogonal matching pursuit(DCS-GOMP)algorithm,is proposed.Compared with other distributed compressed sensing reconstruction algorithms,mainly comparing with the Joint-OMP algorithm,the DCS-GOMP algorithm has a higher probability of successfully reconstructing the signal and the reconstruction error is also low.