The problem of concern here is parameter estimation of harmonics in the presence of multiplicative and additive noise. Cyclic statistics are employed to estimate the frequencies and phases, after which the time series is demodulated and cumulants of the noise processes are estimated. The latter are then supplied to linear or nonlinear cumulant-based algorithms to identify ARMA model parameters for the noises. Cyclic statistics and higher order spectra-based approaches are shown to yield the same frequency estimates. Simulation examples illustrate the algorithms.
Bibliographical noteFunding Information:
Manuscript received April 6, 1994; revised March 15, 1994. Some results of this paper were presented at the 27th CISS conference, John Hopkins Univ., Baltimore, MD, March 1993. This work was supported by ONR Grant no. NOOO14-93-1-0485 The associate editor coordinating the review of this paper and approving it for publication was Dr. Zhi Ding. The authors are with the Department of Electrical Engineering, University of V i a , Charlottesville, VA 22903-2442 USA. IEEE Log Number 9413308.