Exploiting statistical side information to optimize secondary spectrum access
Muhammad Sibtain Hamayun
Date of Issue2016-05-10
School of Electrical and Electronic Engineering
Unlicensed access of bands in the wireless spectra, that have been left under-utilized by the primary (or licensed) users, is the subject that addressed throughout this dissertation. Unlicensed users (or commonly known as secondary users) attempt to efficiently utilize these bands by exploiting the underlying spatial, spectral and temporal opportunities. We demonstrate that partial or complete knowledge of primary activity, transceiver locations and channel gains can be incorporated into the design of secondary access strategies to improve their throughput performance. Surveys have shown that most of the bands auctioned to primary networks are under-utilized. To resolve the issue of spectrum scarcity, the idea of secondary access to the under-utilized bands has gained popularity over the years. Secondary networks are opportunistic in their attempts of acquiring resources, i.e. they search and acquire resources that are not in use by the primary network, and are expected to release them as primary transmissions resume. The quality-of-service of the primary system, and its throughput requirements supersede those of a secondary system. A logical consequence of a secondary system design is the efficiency of grasping and using under-utilized spectral opportunities. Optimizing channel acquisition procedures of SUs for opportunistic access, by incorporating side information, is the essence of the contributions presented throughout this dissertation. Conventional sensing systems prove to be inadequate when it comes to streamlined secondary procedures. Firstly, we optimize a single SU's decision making process that directly impacts the achievable capacity of secondary and primary systems. This decision is based on a threshold, and by incorporating spatial (geographical) side information, the capacity maximization objective is achieved. This dissertation further shows that partial or complete side information allows systems to break away from the conventional methods of channel selection. Typically, channels are defined by the primary activity on them, and are pursued by SUs in the same discrete manner. We show that if side information regarding interference is made available, this discrete method of channel selections becomes suboptimal. By relaxing the discrete constraints of the channel selections, and allowing users to select any band of frequencies, the contention and thus the interference among multiple SUs can be manipulated in a continuous manner. Moving on to a multi-SU system, we shift our focus on the contentions among SUs that play a pivotal role in adversely affecting the transmission efficiency. These contentions result when multiple SUs, in a non-cooperative manner try to access premium frequency resources. Cooperative spectrum access methods minimize contention among SUs by performing orthogonal channel selections, for SUs. The extent of cooperation among SUs may be bounded by system constraints, but the operational conflict-minimization of these SUs can be improved. It has been shown that the thresholds, if carefully selected before sensing begins, can strike an optimal trade-off between premium-channel acquisition and contention minimization. We achieve this goal by modelling the throughput maximization problem to incorporate spatial and channel side information, if made available. Furthermore, a distributed sensing order selection approach is also proposed. Finally, this dissertation also considers the secondary systems that allow SUs to sense multiple channels within a frame, before they can acquire any one of them for transmissions. The order or sequence in which the SUs perform sensing affects the contention among them. We show that if location side information is utilized in the design of sensing orders, the contentions among the SUs is minimized. Without side information, the SU system design has to rely on assumptions regarding the detection outcomes that result in a waste of valuable spatial opportunity. Numerical results and analytical proofs presented throughout this dissertation advocate our thesis of incorporating partial or complete side information for maximal exploitation of the spatial and spectral opportunities left unused by the primary networks.