Improving Resource Management in Dynamic Cognitive Radio Networks through Low-Complexity Channel Selection Algorithms

• Main Author: Zappaterra, Luca
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-2014
• Subjects: Computer Engineering; Computer science; Electrical engineering; Mass communications
• ISBN: 9781303817038; 1303817039

Growing demands for radio-spectrum resources coming from new wireless technologies such as body area networks, self-organizing networks, machine-to-machine communications, as well as from the diffusion of new mobile devices, such as smartphones, tablets, and wearable computers evidenced a shortage of the radio spectrum, posing a restraint for the development of future wireless technologies. Cognitive radio networks (CRNs) represent a solution providing opportunistic access to unlicensed secondary users (SUs) to exploit the spectrum bands not efficiently utilized by the licensed primary users (PUs), denoted as spectrum holes. This dissertation focuses on the resource management for dynamic CRNs, in which the considered channels have rapid and bursty traffic patterns of PUs; SUs seek for temporal absence of primary occupations through channel sensing, immediately transmitting on the spectrum holes identified. Since large parts of the radio spectrum are allocated to PUs with such characteristics, implementing resource management functionalities for SUs to search, select, and exploit the available resources can provide a significant increase of the spectrum utilization, easing the spectrum scarcity problem. This dissertation addresses the key resource management issue of channel selection for dynamic CRNs, which finds and selects the spectrum resources maximizing the SUs' throughput, among channels with heterogeneous PUs' occupations, link qualities, and data transmission rates. Aided by the sensing results and channel statistics, the channel selection can be modeled as an optimal stopping problem. This is typically solved by the dynamic programming method of backward induction, often incurring intractable computational time and storage requirements, or by approximate methods with inaccurate results. Motivated by the lack of general suitable solutions, low-complexity channel selection algorithms are devised, which are characterized by online linear complexity during the selection process and relying on offline pre-computed values, which are re-computed only when the channels' characteristic significantly change. Considering different possible CRNs, the channel selection problem for SUs with single-band transmission capability is formulated and analyzed first, deriving a lightweight backward induction solution. Then, a novel algorithm for the very challenging case of SUs supporting multi-band transmission is devised and analyzed. Finally, a low-complexity solution for SUs employing multi-band transmission with limited power and satisfying the latency requirements of different SUs' applications is proposed. The contributions of this dissertation research lie in the areas of resource management and channel selection for dynamic CRNs. This work is the first to carry out an algorithmic study of the problem of channel selection for dynamic CRNs, which intrinsically contains performance tradeoffs and expensive computations. The developed solutions are novel and hold significant promise to be applied in many emerging wireless applications, thanks to their constrained algorithmic and operational complexities. The superiority of the proposed algorithms over the traditional approaches are demonstrated theoretically and experimentally.