Optimization and experimental characterization of novel measurement methods for wide-band spectrum sensing in cognitive radio applications

Spectrum sensing is a fundamental task in the complex field of cognitive radio systems. It allows a cognitive terminal to scan a frequency span of interest and sense the presence of other users transmitting over it. Many spectrum sensing methods are present in literature and many interesting algorithms have been proposed. Unfortunately, very few methods allow to know the exact boundaries of the user signal, without knowing any channelization of the spectrum of interest. In this context, the paper proposes two novel algorithms, whose purpose is twofold: to keep a low computational burden and to provide information at the most detailed level with respect to the category the algorithms belong to. Tests, executed on simulated and emulated signals, have demonstrated that both algorithms allow reaching a detection probability greater then 95% and a false alarm probability lower then 5% even in scenarios characterized by SNR as low as −10 dB.