Implementation and performance analysis of 2D DFT beamspace ESPRIT

Abstract

The 2D DFT beamspace ESPRIT is a closed form algorithm that provides automatically paired azimuth and elevation angle estimates of multiple sources incident on a uniform rectangular array of antennas. This paper shows how the performance of 2D DFT beamspace ESPRIT can be improved via incorporation of null-steering. Null-steering provides the means for filtering out sources (interferers) in adjacent sectors, while searching for sources in a given spatial sector. Simulation results that document the performance improvement gained by null-steering are presented. The paper also analyzes the statistical performance of 2D DFT beamspace ESPRIT. Asymptotic expressions for the variances of the 2D DFT beamspace ESPRIT DOA estimates are obtained, and verified by computer simulations. The performance analysis results also apply to 2D unitary ESPRIT, the element space counterpart of 2D DFT beamspace ESPRIT.