Energy-Throughput Tradeoff with Optimal Sensing Order in Cognitive Radio Networks

We consider the problem of optimal sensing in a cognitive radio network which has N primary users (PUs) and one secondary user (SU). Our objective is to maximize the throughput that the SU gets for a given budget on the sensing energy. We consider a time-slotted system and the activity of each PU follows a two state (idle and busy) discrete time Markov chain. At the beginning of each time-slot, the SU senses channels in an optimal sensing order (i.e., in the descending order of posterior probability of a channel being idle) until it finds an idle channel. When the SU finds an idle channel, it stops sensing and transmits its packet in the remaining part of the time-slot. At the end of each time-slot, SU updates the posterior probability of N channels. SU spends finite energy to sense and declare a decision (idle or busy) on each channel. Therefore, sensing energy of the SU drains linearly as the number of channels sensed increases. In this work, we study the energy-throughput tradeoff of SU with optimal sensing order. In particular, we find the number of channels that should be sensed in each time-slot such that we achieve maximum throughput with a bound on sensing energy. Our numerical results show the optimum number of channels M, to be sensed in each time-slot for perfect and imperfect sensing.