Periodic sensing in cognitive radios for detecting UMTS/HSDPA based on experimental spectral occupancy statistics

In this paper we study the performance of periodic spectrum sensing in cognitive radios (CR) based on experimentally obtained temporal spectral occupancy statistics of the legacy user (LU). The detection performance of the periodic sensing technique depends on the occupancy statistics of the LU such as the traffic arrival rate and the occupancy duration. The sensing period and the sensing duration requirements also depend on the temporal statistics of the LU in order to achieve a predefined detection probability. The LU services that we consider here are the UMTS voice and the HSDPA data services that operate in the 2100MHz frequency range. The UMTS and HSDPA spectral occupancy behaviors are experimentally analyzed using real world measurements performed by Telefónica I+D in order to find the white spaces or spectrum holes associated with it, which are then used to study the detection performance of the periodic sensing technique. The temporal characteristics such as the occupancy distribution and the approximated occupancy-time plot are presented for the UMTS/HSDPA services based on the experimentally obtained data. The probability of detection is then presented with respect to the sensing period and the duration, based on a temporal spectral occupancy model that we derive from the experimentally obtained occupancy statistics.