Gram-Charlier and generalized probabilistic neural networks based radar target detection in non-Gaussian noise

This study presents the architecture and principle of operation for two classifiers, namely the Gram-Charlier neural network (GCNN) and generalized probabilistic neural network (GPNN), for an application of radar target detection in non-Gaussian noise environments. The GCNN classifier is based on applying the Gram-Charlier series approximation of probability density functions. The GPNN is based on applying the Gram-Charlier series and Parzen's (1962) windowing technique for approximation of density functions. These classifiers are implemented in parallel architectures. Training of these networks using a modified Kohonen training algorithm is presented. GCNN and GPNN have the advantage of requiring very short training times. Performances of these classifiers for radar target detection are evaluated in terms of probability of detection versus signal-to-noise ratio. These classifiers performed respectably well compared to other conventional radar target detectors.<>