Automatic Modulation Classification using Spectral and Statistical features and Artificial Neural Networks

Abstract

Automatic Modulation Classification (AMC) is the process of deciding, based on observations of the received signal, what modulation is being used at the transmitter. It is also becoming increasingly important in cooperative communications, with the advent of the Software-Defined Radio (SDR). Classifying signal types is of high interest in various other application areas such as imaging, communication and control target recognition. Hence, the digital modulation recognizers have critical importance. Ten Digitally modulated signals are generated. Channel conditions have been modeled by simulating AWGN and multipath Rayleigh fading effect. Seven key features have been used to develop the classifier. Higher order QAM signals such as 16QAM, 64QAM and 256 QAM are classified using higher order statistical parameters such as moments and cumulants. Feature based ANN classifier has been developed. Overall classification result obtained for 3dB SNR is more than 97%. The success rate is 99 % (no fading condition) for 5dB SNR value. The developed classifier could classify ten modulated signals under varying channel conditions for SNR as low as -5dB.