Optimal Energy Tradeoff for Active Sensing In Cognitive Radio Networks

We consider a wideband cognitive radio network, where the channel is modeled by doubly block fading in time and frequency. With a frame energy constraint, we propose a joint sensing and probing strategy, called active sensing, to access the best channel for secondary data transmission. By employing the multi-armed bandit problem, we establish the fundamental tradeoff between sensing, probing, and transmitting energies for an average throughput of the secondary user and proactively design the optimal cardinalities of sensing and probing sets of channels. Our design methodology provides a framework to determine the optimal energy tradeoff between exploration and exploitation in wideband cognitive radio networks.