Spectrum sensing in the TV white spaces

Cognitive radio networks operating in the digital television white spaces are of particular interest for their practical applications. In this paper, we review several parametric and nonparametric test statistics commonly used in spectrum sensing. Both single-antenna as well as multiple antenna techniques are considered. For a selected subset of these techniques, an accurate performance assessment is carried out in the presence of a DVB-T primary signal generated using a software-defined realtime transmitter. Sensing performance is assessed both through Monte Carlo simulation and using a real software-defined radio implementation. Different channel profiles are considered. The obtained results show the performance of each algorithm in terms of detection probability under fixed false alarm rate and of Receiver Operating Curve (ROC). Moreover, these results permit to establish clear relationships between the considered algorithms in case of DVB-T primary signals.