Blind calibration for radio interferometry using convex optimization

Radio interferometric arrays as imperfect instruments require calibration to correct for instrumental gain and phase errors in observations. One point of calibration, our focus, is before the time-series of groups of compact receivers are beam-formed. Traditional supervised calibration approaches tune instrumental parameters using prior sky source information. This can prove problematic when information is either insufficiently accurate or missing. We propose to apply the blind (unsupervised) calibration approach for compressed sensing in the calibration of compact phased-arrays. The approach exploits a sparsity description of the sky, using no prior knowledge of actual objects within. Simulations as well as real-data analysis admit that the blind calibration approach is indeed an improvement over existing supervised calibration methods.