Evaluation of Adaptive MAC-Layer Sensing in Realistic Spectrum Occupancy Scenarios

Spectrum scarcity has been discussed as a possible road block that may slow down the growth of wireless communications. Dynamic Spectrum Access (DSA) has been proposed as a possible solution. In this paper, we study the impact of the spectrum occupancy statistics on the MAC-layer sensing performance. Namely, we consider the duty cycle and the distributions of the lengths of time periods during which a channel is busy or idle (ON- and OFF-periods). We use spectrum occupancy data extracted from extensive measurements and generated from accurate models that we previously fitted to our measured traces. We show that if the ON- and OFF-period durations tend to be longer adaptive algorithms that exploit knowledge on these distributions will achieve significant performance gains. If the lengths are shorter the enhancements will be limited. Additionally, also the benefit from investigating more channels changes with the spectrum occupancy statistics. Under more rapidly changing conditions, adding resources for sensing yields more improvements compared to scenarios with longer ON- and OFF-periods.