OPTIMUM AND SUB-OPTIMUM SOURCE LOCALIZATION WITH SENSORS SUBJECT TO RANDOM MOTION.

When an array is used to estimate source location, random motion of the receiving sensors can seriously limit performance. This paper models sensor motion as a flat, bandlimited Gaussian process independent from sensor to sensor Cramer-Rao bounds on errors in time delay estimates are derived and these are converted to bounds on range and bearing errors. The validity of the bandlimited model is examined, and the performance of conventional and optimal processors compared. It is shown that conventional processors can achieve near-optimal performance if the post processing signal-to-noise ratio is sufficiently large and the motion of the sensors sufficiently slow.