Enhanced spectrum sensing for AI-enabled cognitive radio IoT with noise uncertainty

Spectrum sensing plays a major role in Cognitive Radio-based Internet of Things (CR-IoT) for identifying spectrum holes. However, a cooperative CR-IoT approach does not obtain sufficient sensing gain and sum-rate when using the conventional Energy Detection (ED) in an Noise Uncertainty (NU) environment, which may be aggravated under deep fading. To mitigate this problem, we propose an enhanced spectrum sensing technique and sum-rate calculation for Artificial Intelligence (AI)-enabled CR-IoT using the enhanced Kullback-Leibler Divergence (KLD). After a sensing phase, each CR-IoT user performs an enhanced KLD technique using local statistics, which allows us to reduce the required number of samples for reliable sensing. Then, each CR-IoT user sends its local decision to an AI-enabled Coordination Centre (AI-CC) that obtains a decision managing the channel fading state of the CR-IoT users. Finally, this decision is sent to an Fusion Centre (FC) that makes a global decision. The results obtained through simulations show that the proposed enhanced KLD technique achieves detection performance (86%) in comparison with conventional ED technique (69%) and KLD technique (97%) for an NU factor (p = 1.03), number samples (Ns = 30) and channel fading conditions.