The direction deviations of the calibration sources would seriously degrade the calibration precision of the sensor positions.Aiming to this problem

the robust calibration algorithm for estimating the sensor shape against the location deviations is presented under the condition that the prior probability distribution of the direction deviations is available.The robust algorithm is derived based on the signal subspace fitting (SSF) technique and the Bayesian estimation theory frame

and it is numerically implemented via the Newton iteration without optimizing the locations of the calibration sources.Both the theory analysis and simulation experiments validate that the novel algorithm has preferable robustness against the location deviations of the calibration sources and its asymptotic performance can reach the Cram-Rao bound (CRB) under some moderate conditions.