Observed Information Matrices for Multistatic Target and Sensor Field Tracking

Multistatic active target and sensor field tracking in GPS-denied scenarios requires the computation of joint maximum a posteriori estimates of target and sensor field tracks. An alternating directions algorithm, based on a new integral decomposition of the likelihood function of a bistatic range measurement, cycles over two distinct subalgorithms: The first improves the target estimate by estimating the eccentricity of the bistatic ellipse, conditioned on known sensor locations (i.e., the ellipse foci), while the second improves the sensor location estimates, conditioned on known target state (i.e., the ellipse eccentricity). Both subalgorithms are iteratively re-weighted linear-Gaussian Kalman smoothers. Trajectory observed information matrices are given for both target and sensor field estimates. Recursions are derived for efficiently computing the filtered observed information matrices for target and sensor field. The recursively computed, filtered, observed information matrices are proposed as the basis of a sensor management system.