A Self-Organizing Interval Type-2 Fuzzy Neural Network for Radar Emitter Identification

Several classifiers are available for the identification of radar emitter types from their waveform parameters. In particular, these classifiers can be applied to data that is affected by some types of noise. This paper proposes a more efficient classifier, which uses on-line learning and is attractive for real time applications, such as electronic support measures. A self-organizing interval type-2 fuzzy neural network (ST2FNN) is proposed for radar emitter identification. The ST2FNN has both an on-line structure and parameter learning ability. The structural learning includes to add a new rule to an interval type-2 fuzzy neural network (IT2FNN) and to prune an inefficient rule from ITSFNN; and the parameter learning can improve the learning ability of IT2FNN. Then the developed ST2FNN is applied for the radar emitter identification. Simulation results indicate that the porposed ST2FNN can achieve satisfactory classification performance and has a consistent average error deviation level that is lower than that of other neural network classifiers.