Reinforcement Learning Based Transmission Strategy of Cognitive User in IEEE 802.11 based Networks

Abstract: Traditional concept of cognitive radio is the coexistence of primary and secondary user in multiplexed manner. we consider the opportunistic channel access scheme in IEEE 802.11 based networks subject to the interference mitigation scenario. According to the protocol rule and due to the constraint of message passing, secondary user is unaware of the exact state of the primary user. In this paper, we have proposed an online algorithm for the secondary which assist determining a backoff counter or the decision of being idle for utilizing the time/frequency slot unoccupied by the primary user. Proposed algorithm is based on conventional reinforcement learning technique namely Q-Learning. Simulation has been conducted in order to prove the strength of this algorithm and also results have been compared with our contemporary solution of this problem where secondary user is aware of some states of primary user.