The effect of noise variance uncertainty for spectrum sensing in IEEE 802.22.3

The paper analyses the influence of noise variance uncertainty on the performance of an energy detection-based spectrum sensing algorithm thought for the Dynamic Spectrum Access framework. The analysis is based on spectrum acquisitions carried out on surely vacant channels by employing a commercial Software Defined Radio. The effects of noise variance variability is then evaluated in simulation environment by imposing the variations measured in actual power measurements during the experimental campaign lasting 48-hours. The achieved results clearly prove a significant performance degradation in terms of probability of both detection and false alarm, which in turn could provoke the loss of compliance of the spectrum sensing performance with respect to constraints given by international standards, as IEEE 802.22.3. The paper paths the way to further investigations aiming at providing an upper bound to noise power estimation time period in order to keep the noise uncertainty under tolerable limits.