Research On DOA Estimation Algorithm In Massive MIMO System

With its high spatial angular resolution and spectrum efficiency,multiple-input multiple-output(MIMO)is viewed as a key technology for future green wireless communication networks.Direction of arrival(DOA)estimation technology based on massive MIMO provides ultra-high accuracy DOA measurement performance.When the number of antennas tends to large-scale,the DOA estimation of fully digital structures faces a problem of high computational complexity,while the hybrid analog and digital structures face the challenge of how to quickly eliminate phase ambiguity.This thesis focuses these two problems,and the main research content and contributions as follows:1)Three low complexity DOA measurement algorithms,called partitioned subarray auto-correlation covariance matrix combining(PSACC),partitioned subarray autocorrelation angle combining(PSAAC)and partitioned subarray cross-correlation combining(PSCC),are proposed to address the problem of high complexity of the eigenvalue decomposition-based direction measurement methods for massive fully digital receive array.Simulation results demonstrate that when the number of antennas tends to large-scale,the proposed three algorithms can significantly reduce the computational complexity compared to the conventional root multiple signal classification(Root-MUSIC).Among them,PSAAC can reach the corresponding lower bound of variance and PSCC can reach digital CRLB,while PSACC and PSAAC have about 2～4d B performance loss.2)Four algorithms with a single timeslot to eliminate phase ambiguity called maximizing received power(Max-RP),Max-RP plus quadratic interpolation(Max-RPQI),Root-MUSIC Plus Max-RP-QI and two-layer hybrid and digital Root-MUSIC(TLHAD-Root-MUSIC),are proposed to solve the problem of phase ambiguity of DOA estimation for massive hybrid analog and digital structures.Simulation results show that all four proposed methods can eliminate phase ambiguity in a single time slot and have low computational complexity.The performance of the four algorithms in ascending order is as follows: Max-RP,Max-RP-QI,Root-MUSIC Plus Max-RP-QI and TLHADRoot-MUSIC.In addition,Root-MUSIC Plus Max-RP-QI achieves a mixed CRLB in the high signal-to-noise ratios regions with slightly higher computational complexity than Max-RP and Max-RP-QI.Because of the adoption of fully digital structure in part of the TLHAD-Root-MUSIC antenna array structure,its estimate performance is better than those of the first three algorithms and achieves the corresponding CRLB.