In cognitive radio (CR) networks, a static activity model fails to capture the dynamic and time-varying behavior of the licensed or primary users (PUs). In this paper, a distributed scheme is proposed that allows mobile CR users to learn about the activity of the PUs, and disseminate this information to the neighboring nodes that also function as information repositories. In order to guarantee sensing precision and transmission efficiency, the proposed method switches between time-intensive "fine sensing" and quick "normal sensing". Our approach uses the maximum likelihood estimator to learn a...
