The Selective Clustering Energy Detector for Cognitive Radio Networks - Conceptual Design and Experimental Assessment

Spectrum sensing using energy detectors is a promising, simple and flexible approach for cognitive radio networks to detect primary users. In this paper, we present a new energy detector, called Selective Clustering Energy Detector (SCED). This detector determines dynamic thresholds based on clustering and aggregating selected sequences of energy measurements obtained from a software-defined radio. We assess SCED in a series of controlled experiments on real hardware, and compare its performance and reliability to reference cell based energy detectors reported in the literature. We show that SCED surpasses these detectors in all experiments, in particular in scenarios with non-transmission-free reference cells and during the startup phase.