A Decision Tree-based Spectrum Sensing Approach in IoT-based Software-Defined Networks

This paper presents a novel machine learning (ML)-based approach for spectrum sensing (SS) in software-defined networks (SDNs) that coexists with licensed non-orthogonal multiple access (NOMA)-based wireless systems. The proposed solution leverages feature-based SS along with an ML approach to effectively detect idle licensed channels that IoT-based SDNs can opportunistically utilize. The SS framework employs a decision tree (DT) ML classifier, significantly enhancing detection efficiency in Internet-of-things (IoT) environments that typically experience low operating SNR levels. The proposed DT-based approach ensures reliable SS in dynamic IoT settings, where maintaining detection performance in low SNR scenarios is often a significant challenge. The effectiveness of the DT classifier is evaluated based on its detection capabilities. Simulation results demonstrate that the ML-based approach significantly increases the detection probability compared to traditional feature-based methods at low SNR levels, specifically in ranges of -12 dB to 0 dB, offering a robust and scalable solution to improve spectrum utilization in IoT-based SDNs.