Cooperative Spectrum Sensing with Machine Learning Approaches in Cognitive Radio Networks for IoMT Applications

The increasing demand for Internet of Medical Things (loMT) and connected healthcare adds to the cellular systems' limited bandwidth, leading to a scarcity of available fre-quencies. Spectrum sensing in cognitive radio networks (CRNs) can address this challenge by monitoring and accessing under-utilized frequencies, enhancing spectrum efficiency. Compared to several initiatives embracing CRN in generic loT, our work pioneeringly investigates on how CRNs uphold connectivity for the connected healthcare applications. The focus of this paper is on exploiting the robustness of machine learning (ML) algorithms in CRN and employing ML in identifying idle channel states in spectrum sensing for the 10MT application scenarios. Suitable ML algorithms, including SVM, KNN, Decision Tree, Random Forest, and Naive Bayes, are trained to detect primary users and locate vacant bands for optimal allocation, in the presence of secondary users whose data is generated by considering various factors, e.g., received signal strength, path loss, and different fading effects. The system performance is assessed in terms of accuracy, precision, recall, and score for all ML approaches. The numerical results demonstrate a considerable improvement in cooperative spectrum sensing accuracy, with a remarkable 20% reduction in false positives.