Optimal Opportunistic Channel Access in Wireless Communication Networks

Abstract

High transmission rates will be demanded in future wireless communication networks. However, we will soon experience a spectrum scarcity problem since almost all available spectrum has been allocated to various wireless applications. A promising solution to this problem is to significantly improve the spectrum utilization efficiency by using opportunistic channel access (OCA). In the literature, OCA approaches have been developed in two kinds of networks: cognitive radio networks (CRNs) (in which secondary users, which are unlicensed users, may access the spectrum when primary users, which are licensed users, are not active) and wireless networks exploiting time diversity (in which a user may give up its transmission opportunity if its channel is deeply faded). This thesis is focused on optimal OCA in such two networks, with four research components. The first three research components are to achieve optimality in CRNs. The first research component is for the scenario that the statistical information of primary traffic (such as busy/idle probabilities) is known at a secondary user. When a secondary user can sense multiple channels simultaneously but the maximum numbers of channels that can be sensed simultaneously and that can be accessed simultaneously are both limited, we derive optimal strategies to select which channels to sense and which sensed-free channels to access. The second research component is for the scenario that statistical information of primary traffic is unknown and thus needs to be learned during channel sensing and access process (which results in learning loss). When busy/idle states of each channel are independent from one slot to another, we derive secondary channel sensing and access rules with asymptotically finite learning loss or logarithmic learning loss. As an extension of the second research component, the third research component uses another popular channel statistical behavior model: busy/idle states of each channel over time slots follow a Markov chain. We derive a channel sensing and access rule with logarithmic learning loss. The last (but not the least) research component is for optimal distributed OCA that utilizes time-diversity in wireless cooperative networks. Two cases are considered: the case when the source knows channel state information of links from itself to relays and from relays to its destination; and the case when a source knows only channel state information of links from itself to relays. In the two cases, the optimal transmission strategies that maximize the average system throughput are derived theoretically. Our research reveals that time diversity can be exploited in a wireless cooperative network by our proposed strategies.