Kernel fuzzy c-means clustering on energy detection based cooperative spectrum sensing

Cooperation in spectral sensing (SS) offers a fast and reliable detection of primary user (PU) transmission over a frequency spectrum at the expense of increased energy consumption. Since the fusion center (FC) has to handle a large set of data, a cluster based approach, specifically fuzzy c-means clustering (FCM), has been extensively used in energy detection based cooperative spectrum sensing (CSS). However, the performance of FCM degrades at low signal-to-noise ratios (SNR) and in the presence of multiple PUs as energy data patterns at the FC are often found to be non-spherical i.e. overlapping. To address the problem, this work explores the scope of kernel fuzzy c-means (KFCM) on energy detection based CSS through the projection of non-linear input data to a high dimensional feature space. Extensive simulation results are shown to highlight the improved detection of multiple PUs at low SNR with low energy consumption. An improvement in the detection probability by similar to 6.78% and similar to 6.96% at -15 dBW and -20 dBW, respectively, is achieved over the existing FCM method.