Artificial Intelligence-Driven Real-Time Automatic Modulation Classification Scheme for Next-Generation Cellular Networks

Abstract

Automatic modulation classification (AMC) can play an important role in the timely identification of suspicious and unwanted signal activities to enable secure communication in future next-generation cellular networks. Moreover, AMC can detect the modulation scheme without even adding additional overhead in the signal. In this paper, we developed a universal software radio peripheral (USRP) based intelligent AMC system to detect and classify various digital modulation schemes in real-time. For each modulation scheme, we extracted different spectral features for different values of signal-to-noise ratio (SNR) values. Based on the extracted features, we train the neural network to classify the modulation schemes. Experimental results show that we achieve around 97% classification accuracy in real-time as compared to the existing offline classification schemes. Moreover, we also compare the performance of the proposed model with HisarMod2019.1 model in terms of various metrics such as cross-entropy and mean square error. Results clearly demonstrates the efficiency of the proposal for real-time implementation and classification.