Investigation of Neural Network Hardware for Electronic Warfare Applications

Accurate Automation Corporation (AAC) has developed a new artificial neural network processor (NNP) based upon a sparse Multiple Instruction Multiple Data (MIMD) structure which is ideally suited for Electronic Warfare (EW) applications. The proposed research will investigate the use of the AAC NNP for EW applications such as emitter identification which is used in ELINT, ESM, and RWR receivers. This effort will evaluate the NNP's in terms of speed, flexibility, suitability and parallel processing capability. Neural network techniques will be used to classify emitters based upon feature vectors composed of measured operating characteristics such as pulse repetition frequency (PRF), pulse repetition interval, frequency, modulations, etc. Once developed the techniques will be implemented in the AAC NNP and compared to results obtained previously for the Intel Electrically Trainable Artificial Neural Network (ETANN) chip and others reported in the literature.