A New Frequency Estimation Approach to DFH Signal with Periodogram and ARMA

This paper is intended on the estimation of the hopping frequencies of the differential frequency hopping (DFH) signals in white and colored Gaussian noise, respectively. A method based on the combination of the periodogram and the autoregressive moving average (ARMA) modeling-total least squares (TLS) algorithm, is proposed. The spectrum of the signal is firstly evaluated using the periodogram. The frequency estimation can be directly obtained by peak-value-searching of the spectrum, if the frequency points are far apart spaced. When they are closely spaced the singular value decomposition (SVD)-TLS algorithm is employed. Fourth-order cumulants are used in accordance with the ARMA model for colored Gaussian noise. This method enjoys the advantages of both algorithms, i.e. low computational complexity, simple implementation and high accuracy. It also avoids false frequency point estimations which might appear when the SVD-TLS algorithm is employed only. Simulation results verify the feasibility and effectiveness of the method