Bayesian subspace estimation for beamforming and RFI cancellation using deflation technique

Abstract

The output performance of a radio telescope is always suppressed severely by strong interferences. To solve this problem, a new Bayesian subspace estimation algorithm is proposed for beamforming and RFI cancellation. Assuming the steering vectors of interferences follow complex Gaussian distribution, the proposed algorithm recursively estimates these steering vectors based on Bayesian principle and subspace deflation technique, and then the span of interference subspace is realized. Numerical simulations show that, compared with eigenvalue decomposition (EVD) and fast approximated power iteration (FAPI), the proposed algorithm can accomplish a lower subspace estimation error and a better performance in beamforming and interference cancellation, especially when weak signals are considered.