Automatic modulation classification and performance analysis under Awgn using machine learning in cognitive radio

In domain of cognitive radio, capability to automatically segregate various modulations under varying signal- to- noise ratios (SNRs) plays a crucial role in ensuring reliable communication. The research focuses on the development and implementation of an efficient machine learning-based approach for automatic modulation classification (AMC) in cognitive radio systems. The project involves the analysis of different modulation schemes such as quadrature amplitude modulation 16(QAM16), quadrature amplitude modulation 64(QAM64), and quadrature phase shift keying (QPSK) among others, considering a large range of SNRs. The intended methodology leverages advanced machine learning algorithms, including but not limited to deep learning models decision tree classifier, k-nearest neighbor (KNN), and random forest classifiers for accurately classifying the received signals. By utilizing a diverse dataset generated through simulations and practical experiments, the working performance of the classifiers is thoroughly estimated. The classification accuracy is carefully assessed for different SNR levels to determine the robustness of the classification system under challenging environmental conditions. Additionally, this study investigates the impact of varying SNRs on the modulation classification performance, aiming to provide insights into the system’s resilience to noise and interference. The project’s findings contribute to the enhancement of cognitive radio systems, facilitating their adaptability and efficiency in dynamically allocating spectrum resources. The results shows that the significance of leveraging machine learning algorithm in ensuring reliable and adaptive communication in cognitive radio networks.