Recursive direction finding in the presence of sensor array uncertainties

An algorithm which recursively computes the maximum likelihood (ML) estimates of the arrival directions of multiple signals in the presence of sensor array uncertainties is presented. It is assumed that a priori estimates of the signal arrival directions and sensor locations and intrinsic phases are available and that they are reliably described by Gaussian random variables with a given means and covariances. The algorithm combines the a priori information together with new data collected by the array and then produces updated estimates and covariances of both the signal arrival directions and the unknown sensor parameters. Therefore, not only does this recursive procedure generate optimal estimates of the signal locations, but it also performs sensor self-localization and calibration.<>