Effective energy detection using machine learning techniques for cooperative sensing in cognitive radio networks

Effective energy detection using machine learning techniques for cooperative sensing in cognitive radio networks

연구 분야: Infrastructure

학회: Multimedia Tools and Applications

In this study, the focus is on improving spectrum utilization and reducing interference in wireless communications through Cognitive Radio Networks (CRNs). The approach involves the integration of machine learning and Cooperative Spectrum Sensing (CSS) in a centralized CR network architecture. The objective is to effectively differentiate between signal-plus-noise bands and noise-only bands. MATLAB simulations were conducted to evaluate the proposed model under varying fading and non-fading scenarios, considering a network configuration with three primary users and five secondary users. The results highlight a notable overall accuracy of 86.6%. Additionally, the calculation of Cohen's Kappa coefficient yielded a value of 0.732, indicating substantial agreement and classification reliability in the model. The efficacy of this approach was further demonstrated through the successful training of a Convolutional Neural Network (CNN) model. This training utilized 1000 samples, allocated into 70% for training, 15% for testing, and 15% for validation, spanning 2000 epochs and 8000 iterations. In the simulated deployment scenario covering a geographical area of 120 km × 120 km, cooperative spectrum sensing performance was analysed with varying secondary user counts (N = 5 and N = 10). These quantitative evaluations underscore the significant potential of the model in enhancing spectrum sensing capabilities, promising more efficient and reliable utilization of the wireless spectrum in CRNs.