A novel hybrid algorithm for two-dimensional direction-of-arrival estimation in massive MIMO systems

Abstract

Massive Multiple-Input Multiple-Output (mMIMO) systems have been significantly exploited to enhance mobile network capacity. Increasing the number of antennas provided higher direction-of-arrival (DOA) estimation accuracy, especially when utilizing a two-dimensional multi-signal classification algorithm (2D-MUSIC). However, this algorithm suffers from the heavy computational complexity imposed by the 2D angle search. This paper proposes a novel hybrid algorithm called the 2D-MM, which combines the strengths of the 2D-MUSIC algorithm with the 2D-Minimum Variance Distortionless Response (2D-MVDR) algorithm to reduce the computational complexity using a fast new DOAs search strategy. To be more precise, the new search strategy involves two stages: the 2D-MVDR stage detects the approximate DOAs within a whole angular field of view using as few antennas as possible and snapshots of received data. Then, the 2D-MUSIC stage accurately determines the number and locations (azimuths and elevations) of those sources within the angular sectors in which they appear. Furthermore, the choice of utilizing a uniform circular array (UCA) configuration for representing mMIMO systems is enhanced by introducing a new geometric structure featuring two concentric rings. The simulation results indicate that confining the spectral peak searching process into octant areas with an angular sector spanning π⁄4 instead of the total angular field of view (0–2π) would avoid searching empty areas and then reduce the computational complexity, where the improvement percentage in the proposed algorithm's running time was approaching 80 % when utilized with mMIMO arrays. In contrast, the proposed algorithm proves effective for mMIMO systems, irrespective of the increased number of antennas utilized.