A Stochastic CRB for Non-unitary Beam-space Transformations and its Application to Optimal Steering Angle Design

Many of modern radar systems employ antenna arrays with thousands of anetnnas. These antennas are usually grouped into a few subarrays or beams through the process of beamspace transformation, to estimate the angle of a target. In this letter, we consider a simple but pragmatic beamspace transformation that only allows beam steering, and present a technique of finding optimal steering angles that minimize the Cramér-Rao bound (CRB) for angle estimation. The CRB and mean squared error (MSE) of the angle estimate using the resulting optimal steering angles are compared with those using the widely used DFT-based beam steering.