Experimental Assessment of Spectrum Sensing Under Time-Varying Channels for Cognitive Radio

Cognitive radio (CR) networks rely on spectrum sensing (SS) tech-niques to determine the availability of unused frequency bands. However, their experimental evaluations on a high-mobile wireless systems requiring complex and costly on-site measurements are still a challenging task. In this letter, we evaluate a SS detection under such scenarios by emulating time-varying multiantenna channels on a software-defined radio (SDR) platform. In doing so, the unexplored detection performance of newly conceived detectors can be assessed in mobile environments such as the vehicular networks. The proposed platform is composed of universal software radio peripheral (USRP) boards providing realistic radio frequency (RF) channels, where a mean-to-square extreme eigenvalue (MSEE) detector is evaluated through exhaustive measurement campaigns. The proposed testbed is able to reproduce the simulation results allowing us to evaluate the detection performance under challenging scenarios. For instance, a degradation of only 8.3% in the detection performance has been found when reducing 50% the sensing period for higher Doppler frequencies, whereas this detection is degraded by up to 6.6% for short sensing periods despite the higher Doppler effect. These findings are interesting for the conception of new sensing strategies in time-varying scenarios.