Tracking direction-of-arrival for wireless communication with multiple antennas

The goal of this contribution is to bind together results available for radar systems for direction-of-arrival (DOA) tracking and wireless communications with multiple antennas. We show that for accurate tracking of DOA the MUSIC-like methods based on inversion of the channel snapshot covariance matrix are suboptimal. In our approach we apply an alternative algorithm developed by H. Gu (2002) that uses the instant channel snapshot to update the DOA estimates directly. We improve on this algorithm by introducing online data association capabilities based on linear extrapolation that leads to better tracking performance and lower computational complexity. The performance is evaluated using simulated channels as well as real wideband multiple-input-multiple-output (MIMO) measurements at 2 GHz. The obtained estimates confirm previous results obtained with alternative techniques but provide more accurate results at lower computational complexity.