Modeling time-varying aggregate interference from cognitive radios and implications on primary exclusive zone design

Accurately characterizing the time-varying nature of the aggregate interference from secondary users (SUs) is essential in ensuring a successful deployment of a cognitive radio (CR) network. This requires characterizing the probability distribution and the time-varying nature of the aggregate interference from multiple SUs. We show that it is well modeled as a shifted lognormal random process, and is more accurate than the lognormal, Gaussian, and symmetric truncated stable models considered in the literature even for a relatively dense deployment of SUs. Our model accounts for the effect of imperfect spectrum sensing, which depends on path loss, shadowing, and fading of the link from primary transmitter to the SU, and the randomness in the number of SUs and their locations. It also allows for both interweave and underlay modes of CR operation. We also demonstrate the relevance of the proposed analytically tractable model in the design of the primary exclusive zone.