Automatic Digital Modulation Classification using Extreme Learning Machine in Frequency Offset

Abstract

One of the major problems in non-cooperative and intelligent communication systems is the determination of the type of modulation of the received signal. The problem becomes more challenging when there are synchronization errors such as frequency offset and timing offset particularly in real scenarios. In this study, we propose a new automatic modulation classifier that can determine the digital modulation of signal during the frequency offset taking place in the receiver. Performance analysis of the proposed system is simulated for different signal to noise ratios. The obtained results show low calculation complexity and thus be highly successful when compared with traditional classifiers in the low SNR values.