Adaptive generation of decision functions for classification of digitally modulated signals

The performance of an adaptive technique for the generation of decision functions used in classification of several types of digitally modulated signals-binary amplitude-shift keying (BASK), binary phase-shift keying (BPSK), quaternary phase-shift keying (QPSK), and binary frequency-shift keying (BFSK)-is discussed. A computer simulation is performed to measure the properties of the classifier which is trained with one signal from each class. The method correctly identifies all the signals considered during this experiment. The most interesting result from a small number of trials is considered to be that a novel feature, useful for the identification of the number of phase states of BPSK and QPSK signals, is seen to provide adequate information identification at SNRs down to 5 dB. This feature is the existence of narrowband energy near M times the carrier frequency for M -ary PSK signals