An Empirical Study on the Performance of a Spectrum Sensing Scheme for Cognitive Radio

This paper describes the implementation and the experimental performance of a spectrum sensing scheme, based on a Constant False Alarm Rate (CFAR) algorithm, for cognitive radio in TV white space (TVWS) operation. We use various signal profiles, representing an incumbent or primary user (PU) in the channel, along with a physical testbed to experimentally measure the detection of a PU signal over the following parameters: (1) sample size or frequency resolution, (2) noise estimate and (3) time variability ("burstiness" or "duty cycle") of the PU signal, including using actual IEEE802.11 packets for the primary user. The measured results validate the behavior of the CFAR energy detector scheme on the number of samples and noise value estimate used. We also note that the accuracy of the spectrum sensing scheme, which depends on the signal energy exceeding a threshold according to the CFAR algorithm, degrades at smaller duty cycles of the PU for a given detection window duration and resolution bandwidth.