Statistically optimal self-calibration of regular imaging arrays

Many imaging arrays have a regular sensor configuration. This regularity can be exploited for self-calibration of the array. In this paper, we introduce a new self-calibration method for regular arrays based on weighted alternating least squares (WALS) optimization that appears to be statistically efficient and does not impose requirements on the source structure or on pre-calibration of the array. We show results from Monte Carlo simulations indicating that the proposed method already attains the Cramer-Rao bound (CRB) at very low SNR and produces unbiased results. Our simulations also indicate that the approach most commonly used in the literature does not attain the CRB at high SNR and produces biased results at low SNR.