RECURSIVE ESTIMATION OF FREQUENCIES AND FREQUENCY RATES OF MULTIPLE CISOIDS IN NOISE

A recursive algorithm for simultaneous estimation and tracking of inst antaneous frequencies and instantaneous frequency rates-of-change for signals that consist of multiple narrow-band components in noise is pr oposed and studied. The algorithm recursively separates the signal to individual components and uses estimated phase differences for updatin g the instantaneous frequency and frequency rate of each component. Th e main advantages of the proposed algorithm over frequencies-only trac king algorithms known in literature include the zero asymptotic bias ( zero tracking delay) in estimating of the instantaneous frequencies of linear FM (chirp) signals and more accurate tracking of frequencies t hat cross each other. Performance of the algorithm is studied by means of a linear filter approximation technique and derived results are co mpared with the appropriate (posterior) Cramer-Rao bound. Superior per formance of the algorithm is illustrated by computer simulations. (C) 1997 Elsevier Science B.V.