Perspective methods of radar pulse classification

This paper deals with perspective methods of radar pulses classification. Ability to classify pulses is required by ELINT in order to deinterleave signals and unambiguously identify the source of analyzed signal. Experiments has shown that the best classification methods are based on auto-correlation, cross-correlation, and frequency spectrum analysis. Some other pulse feature are incorporated for the purpose of computation acceleration. The most useful features are the pulse width, carrier frequency, instantaneous frequency variations and instantaneous amplitude variations. Testing has been performed using real signals, emitted by different radars in L, S, C, X bands. A new program for a pulse classification has been written in MATLAB. The program automatically computes key features of an unknown pulse and compares them to data stored in a database. Acquired results proves, that this kind of analysis and classification is usable in real applications.