On estimation of instantaneous bandwidth

In many applications, the instantaneous frequency of the observed signal describes some physical phenomenon of interest. The instantaneous frequency can be interpreted as the mean frequency at a particular instant of time. Then it is natural to supplement any measure of the instantaneous frequency by the estimate of its instantaneous spread or bandwidth. Assuming a polynomial model of log-envelope of an AM-FM signal, the paper first considers the maximum likelihood (ML) estimation and Cramer-Rao bounds for instantaneous bandwidth (IB). Then two simpler algorithms for IB estimation are proposed: the log-envelope difference algorithm and the LS method.