Classification of digitally modulated signals using cross time-frequency distribution

Abstract

Signal classification is important in a wide range of application and various researches had been carried out. This is a challenging task especially in the non-cooperative environment where there is no information of the received signal available. Therefore, time-frequency analysis is applied to extract the signal parameters received off the air. The estimated parameters are then used as the input to the classifier at which the accuracy of the classification will depends greatly on the parameter estimations. This paper proposed the cross time-frequency distribution (XFTD) to classify various type of digitally modulated signals such as ASK, FSK, PSK and QAM signals. It is shown that the XTFD is capable to give above 90% classification accuracy at a minimum SNR of -2 dB. The XTFD is superior to noise and is capable to cover a wide range of signals including digitally phase modulated signals.