Fast and accurate radio interferometric imaging using krylov subspaces

We propose a fast iterative method for image formation in Radio Astronomy (RA). We formulate the image formation problem as a maximum likelihood estimation problem to estimate the image pixel powers via array covariance measurements. We use an iterative solution method based on projections onto Krylov subspaces and exploit the sample covariance error estimate via discrepancy principle as the stopping criterion. We propose to regularize the ill-posed imaging problem based on a Bayesian framework using MVDR beamformed data applied as a right preconditioner to the system matrix. We compare the proposed method with the state-of-the-art sparse sensing methods and show that the proposed method obtains comparably accurate solutions with a significant reduction in computation.