Neural blind separation for electromagnetic source localization and assessment

We present a possible approach to electromagnetic source localization using a hybrid blind separation-minimum search inversion algorithm. The total electrical field versus time emitted by the working antennas, located at different and unknown geographical positions, is used to reconstruct each separate contribution via a suitable neural network technique. When the emitted electric field and the related base frequencies have been separated for each emitting antenna, the unknown location of each emitter, is determined with a minimum-search numerical technique. The theory presented has been applied with success to a practical problem dealing with amplitude-modulated radio-transmissions.