FIR PREFILTERING IMPROVES PRONYS METHOD

Prony's method is a technique for determining the parameters of sinusoidal signals. It is known to be very sensitive to measurement errors in the observed signal samples. In this correspondence, we present two new prefiltering methods that improve the performance of Prony's method. The simple prefiltering operation reduces the effects of noise in the data without violating the fundamental property of Prony's method. In the first method the prefilter is synthesized using prior knowledge regarding the approximate regions in the frequency domain where the signals are known to lie. In the second method no prior knowledge is necessary. The prefilter is iteratively computed from the data samples themselves. Simulation results are given. Connections to other linear prediction methods are also indicated.