Research On DOA Estimation Algorithm Based On Compressed Senseing In Massive MIMO Systems

Massive multiple-input multiple-output(MIMO)for 5G New Radio(NR)has important research significance in mobile communications.Massive MIMO deploys hundreds of antennas at the base station on the transceiver,which has a good spatial multiplexing gain and greatly improves the spatial resolution of the array elements.Besides,a large number of antennas,massive MIMO can accurately estimate the arrival angles based on the phase difference of electromagnetic waves reaching different array elements.Direction-of-arrival(DOA),as a hot area of research in array signal processing,has important applications in social life,the economy,and other fields.At the same time,the accurate angle information obtained by DOA estimation plays a key role in distinguishing target users from interfering users and accurately transmitting effective information in a mobile communication system.Therefore,the DOA estimation problem in massive MIMO systems is of great research value and application significance.Compressed sensing algorithm is a type of algorithm that has been widely studied in recent years.It has good applicability and can reconstruct the target signal of interest with high precision.We introduce compressed sensing theory,and the research on DOA estimation algorithms in massive MIMO systems is studied.We effectively utilize the signal structure and spatial distribution information and optimize the cost function of the algorithm,which significantly improves the performance of the sparse reconstruction algorithm.We propose a high-precision DOA estimation algorithm based on improved block sparse Bayesian learning.Furthermore,this dissertation uses a deterministic approximation method to avoid the direct solution of the posterior probability distribution,and proposes an improved sparse Bayesian learning DOA estimation algorithm based on variational inference.After the verification of the simulation experiment and the analysis of the experimental results,it can be demonstrated that the proposed algorithm can still show great performance in the harsh experiment environment.