Clustered cooperative spectrum sensing based on SNR-means

Hard clustering methods, represented by K-Means, exhibited weak adaptability to complex electromagnetic environments, and their sensing performance was susceptible to the impact of interference. To address this issue, a clustered cooperative spectrum sensing method based on SNR-Means was proposed. This method abandoned the spatial layout-based clustering approach and instead started from the electromagnetic spectrum characteristics of the sensing environment. Using the minimum sum of squared errors of the received signal-to-noise ratio among sensing nodes as the objective function, the sensing nodes was partitioned into multiple clusters with equidistant signal-to-noise ratios. Additionally, factors such as spatial layout, connectivity, and local sensing performance were considered to reasonably designate cluster heads. Based on the local spectrum sensing capability of the nodes, weights were appropriately assigned, and the sensing observation data from each node was weighted and fused, thereby improving spectrum sensing performance. The proposed method significantly reduces the impact of the interference on the results of spectrum sensing and can be effectively applied to spectrum sensing in complex electromagnetic environments.