Radar vericilerinin kimliklendirilmesi için yeni metotlar

In this thesis new methods are introduced for radar emitter identification. Radar emitter identification is the process of identifying surrounding threat emitters in electronic warfare environments. A method is developed for deinterleaving of radar pulse sequences. For this purpose, first clustering performances of two self-organizing neural networks, namely SOM and Fuzzy ART are evaluated. Then, a pulse amplitude tracking algorithm is proposed for dynamically varying signal environments wherein radar parameters can change abruptly. Simulation results show that the proposed algorithm can successfully deinterleave radar emitters that have agile pulse parameters. Another method is developed for the recognition of pulse repetition interval modulation patterns. The method is based upon new features extracted from multiresolution wavelet decompositions of different types of pulse repetition interval modulation sequences. Simulation results show that recognition performance of the proposed features outperform conventional histogram based methods in both accuracy and computation time.