A testbed for automatic modulation recognition using artificial neural networks

Abstract

Automatic modulation recognition, the identification of the modulation scheme used to encode an unknown radio transmission, is an important component of electronic warfare communications systems. Existing technology is able to classify reliably (success rate /spl ges/90%) only at signal-to-noise ratios (SNRs) above 10 dB. In this paper, an artificial neural network is developed to classify signals with SNRs as low as 5 dB. Preprocessing steps and network performance on an initial test set are described.