Investigation of automatic analog modulation classification algorithms

In this paper, feature based modulation classifiers are investigated for AM, DSB, LSB, USB and FM analog modulations methods. Instantaneous phase, magnitude and spectrum of the signals to be classified are used for the calculation of the key features. At the decision step decision tree, minimum distance classifier and support vector machines are used. Then the performance of the developed classifiers is assessed through computer simulations and the decision tree classifier is realized on an USRP software radio platform.