The performance of feature-based classification of digital modulations under varying SNR and fading channel conditions

Abstract

Feature-based classification is a method used for automatic modulation classification of communication signals. This method requires extraction of various features from a signal. One of the approaches for feature extraction is the use of instantaneous amplitude, phase and frequency. The performance of these features under different SNR conditions is well described in literature [1]-[2], but the effect of fading on the extracted features has received little investigation. In this paper we investigate the quality of the features in a static multipath Rayleigh fading channel under varying channel conditions. It was found that most feature values are distinguishable for the different digital signals in a flat fading channel. This work shows the behaviour of these features in a fading channel under deteriorating conditions and suggests ways to reduce the effect of the fading on the classification of the signals