|dc.description.abstract||Under the contemporary spectrum usage regulations, radio frequency bands are allocated statically to licensed users in a large geographical area and over a long period of time. Recent investigations have revealed that such static spectrum allocation has led to very poor usage of the overall spectrum. Cognitive radio has emerged as a new communication paradigm to improve the utilization of the radio spectrum. It is defined as an intelligent wireless communication system that allows coexistence of unlicensed users with the licensed ones as long as the perceived interference at the licensed user is capped below some acceptable level. In addition, the users in this system adopt efficient communication protocols to enhance spectral efficiency.
We employ cooperative mechanisms wherein multiple users cooperate in order to accomplish the following tasks:
1) Cooperative spectrum sensing: In this task, the licensed users do not actively engage. Instead, the unlicensed users passively monitor the activity of the licensed users and transmit only during their absence. 2) Cooperative spectrum management: The licensed and unlicensed users can benefit from cooperation with each other, e.g., they can assist each other in transmission via relaying. In this fashion, they can save power or bandwidth and therefore, the whole network can accommodate more users.
In the first part of this thesis, we focus on cooperative spectrum sensing. We first study the performance of the optimal distributed detectors as the number of samples increases and identify the conditions under which the highest or lowest asymptotic performance is achieved. For each condition, we study several suboptimal detectors and obtain novel asymptotic expressions for their performance. We then consider distributed detection of an Orthogonal Frequency-Division Multiplexing (OFDM) signal source. We propose different optimal and suboptimal frequency-domain detectors and derive closed form expressions for their performance. These frequency-domain detectors, despite their lower computational complexity, outperform the state-of-the-art time-domain detectors. Finally, we consider distributed spectrum sensing in mixture-Nakagami fading channels. We propose several novel detectors that significantly outperform the traditional detectors. In all these cases, we prove that the suboptimal detectors are asymptotically optimal, i.e., their performance converges to the Uniformly Most Powerful (UMP) tests as the number of samples increases.
In the second part of the thesis, we focus on cooperative spectrum management. We study the problem of cooperative relay selection and power allocation and determine the conditions, in terms of channel gains and network geometry, under which such cooperation leads to an increase in rate, or a reduction in power and bandwidth usage. Lastly, we propose cooperative protocols that exploit these results and greatly enhance spectrum efficiency.||en