Model-free Adaptive Rate Selection in Cognitive Radio Links

TitleModel-free Adaptive Rate Selection in Cognitive Radio Links
Publication TypeConference Paper
Year of Publication2013
AuthorsÁ. Gonzalo, and J. Pérez
Conference Name5th International Conference on Mobile Networks and Management (MONAMI)
Month PublishedSeptember
PublisherSpringer-Verlag. LNICST series
Conference LocationCork, Ireland
KeywordsCognitive radio (CR), n-armed bandit problem, rate control, re- inforcement learning (RL)
AbstractIn this work we address the rate adaptation problem of a cognitive radio (CR) link in time-variant fading channels. Every time the primary users (PU) liberate the channel the secondary user (SU) selects a transmission rate (from a finite number of available rates) and begins the transmission of fixed sized packets until a licensed user reclaims the channel back. After each transmission episode the number of successfully transmitted packets is used by the SU to update its optimal rate selection ahead of the next episode. The problem is formulated as an n-armed bandit problem and it is solved by means of a Monte Carlo control algorithm.
PDF version: