Emitter recognition using fuzzy adaptation of ARTMAP neural networks

Emitter Recognition is the problem of classifying the radar type from intercepted radar signals. This capability is crucial for classifying approaching enemy ships and aircrafts. The sensed parameters may vary from their actual or reported values because of manmade variations in the form of agility or staggering. Another cause of variation could be dispersion because of atmospheric effects and equipment noise. Associating the measured radar parameter set with a known sighting is a pattern recognition problem in a multi-dimensional space. Several authors have attacked the problem with various data association tools with different merits and de-merits. Most of them are marred by the massive computing power required and unrealistically large training data requirements. In this paper a simple but elegant technique is proposed to solve the above problem using wellestablished framework of Fuzzy Logic and Neural Networks.