Distributed Algorithm to Learn OSA Channels Availability and Enhance the Transmission Rate of Secondary Users

This manuscript addresses the problem of the Opportunistic Spectrum Access (OSA) in Cognitive Radio Networks (CRNs). In a CRN, the main aim of a secondary user (SU), i.e. an opportunistic user willing to transmit over unused licensed channels, is to learn the availability of channels and find the most often vacant in order to increase his transmission rate. Recent works use multi-armed bandit (MAB) framework to solve the OSA problem for a single SU. However, in this paper a novel decentralized policy is proposed for the multi-user case to estimate collectively the availability of channels using sensing decisions. This policy does not require any cooperation among the secondary users where two sets of SUs are considered: priority (SU
_p
) and ordinary (SU
_o
) users. Our policy ensures that the users SU
_p
often access the best channels while the users SU
_o
access the almost-best ones. Moreover, this policy ensures a fairness among the priority users SU
_p
as well as the ordinary users SU
_o
by giving a uniform access to available channels. In this manuscript, we also propose a novel transmission technique to enhance the transmission rate as well as to extend the transmission range of the ordinary secondary users by requesting the priority ones to play as cooperative relays. In our transmission technique, we consider that the priority users SU
_p
are the nearest to the primary base station. Simulation results show that the users, i.e. priority and ordinary users, access uniformly the channels using our learning policy. We also show that ordinary users achieve higher transmission rate using the proposed transmission technique.