Comprehensive analysis of frequency estimation methods for electronic support systems

In electronic warfare systems, the correct estimation of the threat radar frequency by the electronic support system is of great importance for the identification of the radar and the realization of electronic attacks. The methods used for frequency estimation are divided into two as frequency and time domain methods. Fast Fourier transform (FFT) in the frequency domain is a widely used method. However, when the intermediate frequency (IF) of the radar signal is not an exact multiple of the FFT resolution, the correct frequency value cannot be obtained in the FFT computation. For this reason, Jain, corrected Jain, Quinn, improved Quinn, Jacobsen, Macleod, Ding, Voglewede, MI, Candan, rectangular window-based interpolation and Hanning window-based interpolation methods have been suggested in the literature to improve the frequency obtained as a result of FFT. In the time domain, Digital Instantaneous Frequency Measurement (DIFM), estimation of signal parameters via rotation invariance (ESPRIT), and Multiple Signal Classification (MUSIC) methods are commonly used to estimate threat radar frequency. However, these methods cannot accurately predict the frequencies of synchronous signals. Prony's and Pisarenko's methods have been proposed in the literature to improve frequency estimation in simultaneous signals. In this thesis, the above-mentioned time and frequency domain methods used in frequency estimation are analyzed in the presence of two different noises such as Gaussian and Laplace in a bandwidth varying between 100MHz and 1000MHz and their performances are observed. As a result, findings on which frequency estimation method can be used in which conditions are presented.