Elastic Fusion Dual-stage Spectrum Sensing for Random PU Accessing

Abstract

According to the channel usage right, licensed users (PU) can leave or access the license channel randomly. Then, the sensing slot may not contain the PU signal in all data samples which affect the predetermined threshold of traditional spectrum sensing techniques which is generated according to the available channel condition. Then, the traditional techniques present poor detection performance when the PU leaves or accesses the channel during the sensing slot. Once the late access behavior is concerned, the multi-slot solution is adopted with the spectrum sensing algorithm. Then, the data fusion rule becomes the major function to declare the channel status. Once the multi-slots spectrum sensing methods did not take, their fusion rules present poor detection performance when PU leaves the channel early. In this paper, we propose a new spectrum sensing - elastic fusion dual-stage spectrum sensing (EFDS) - with an elastic fusion rule to address the random access behavior issue. The simulation results show that EFDS presents reliable detection performance under normal usage and noticeably outperforms the detection performance of existing techniques under random usage behaviors.