Subspace approximation for adaptive multichannel radar filtering

In this paper we consider a subspace approximation tailored to adaptive airborne radar. Motivation for this research includes the need for reduced computational burden and approaches for practical implementation. Measured radar data only approximately satisfies the statistical assumptions intrinsic to the adaptive processor. Hence, approximate numerical methods for adaptive weight computation may successfully be used in place of exact methods. We propose a numerical procedure based on partial bi-diagonalization of the interference covariance matrix, coupled with a preconditioned conjugate gradient iterative method, to approximate the dominant subspace and construct the adaptive weights. Through example, we show the potential of this method for adaptive radar.