Real-time spectrum occupancy monitoring using a probabilistic model

The scarcity of the radio spectrum has motivated a search for more optimal and efficient spectrum management methods. One of these methods is spectrum sharing, which multiplies the number of devices that can use this resource without causing harmful interference to licensees. Spectrum sharing requires spectrum scanning to gain awareness of the spectrum occupancy patterns and decide how to allocate access to this resource. This process has been traditionally done by sensing the channel to determine its state, occupied or empty, and then using frequentist inference to estimate the channel occupancy. However, frequentist inference does not handle uncertainty and does not take into account the probabilities of false alarm and detection when estimating the channel occupancy rate. On the other hand, Bayesian inference can handle uncertainty by considering the impact of these parameters on spectrum sensing results. Additionally, it is possible to include previous knowledge into the construction of Bayesian models to learn and make decision under uncertainty. In this paper, we propose a spectrum scanning method, Bayesian inference, to estimate the channel occupancy rate. One advantage of this method is that it takes into consideration the probabilities of false alarm and detection of the spectrum sensor. This feature makes the estimation of the channel occupancy rate more accurate.