Darbe tekrarlama aralığı tanıması ve gauss olmayan gürültü altında gerçek zamanlı tespiti

In this thesis study, feature extraction related with Pulse Repetition Interval (PRI) modulation recognition, as one of the essential processes in Electronic Support (ES) receivers, and the detection of PRI modulated signals under non-Gaussian noise is investigated. The discriminative feature extraction methods in the thesis are primarily based on autocorrelation and power spectral density similar to the literature. The main contribution is the recovery of rectangular pulse onset under α-Stable noise environment which exhibits impulsive non-Gaussian behavior. The first proposed approach is to utilize robust estimators which give limited time of arrival estimation performance depending on the impulsiveness of the channel noise and used robust filter length. As another novel contribution, time-varying cumulative summation (CUSUM) based quickest detection (QD) method is adapted to detect pulse onset with minimum time delay. It is shown that raw PRI modulated waveforms under non-Gaussian noise can be tracked to detect rising edge of the rectangular radar pulses with minimum time delay depending on the selection of proper threshold which is directly related with probability of false alarm.