Memorial K-means clustering for cooperative spectrum sensing in cognitive wireless sensor networks at low SNR regimes

Purpose The spectrum resources are becoming increasingly scarce and underutilized, and cooperative spectrum sensing (CSS) in cognitive wireless sensor networks (CWSNs) offers many solutions with good results, but this paper aims to address the significant issue of CSS in the context of low signal-to-noise ratio (SNR). Design/methodology/approach This study proposes Pearson Correlation Coefficient (PCC) to obtain value feature values under the Rayleigh channel model, which are then used for Memorial K-means Clustering (MKC) analysis of CSS in CWSNs at low SNR regimes. In addition, MKC algorithm is used for training and converted it into supervised model. Findings A series of numerical simulation results demonstrate that the correctness and effectiveness of the proposed MKC, especially the detection and false alarm probabilities in a low SNR condition. The detection probability is increased by 5%–12% at low SNR compared with other methods. Originality/value The MKC algorithm can reduce the impact of randomness on the clustering centers for multiple groups, which combined with PCC can effectively reduce the influence of noise at low SNR, and the unsupervised transformed model effectively reducing the complexity of re-discrimination.