Detection of harmonic signals

Abstract

Harmonic signals are often produced by man-made sources. Their detection has traditionally been accomplished by using Fourier methods. An alternative approach that uses a maximum likelihood function is presented in this paper. The power of this approach comes from modelling the received signal as a family of harmonic signals. This effectively reduces the problem to a one-dimensional search for the fundamental frequency. The coefficients for the various harmonic frequencies can then be found using the Moore-Penrose inverse.