Cognitive radio networks: Resource allocation and effect of end-user behavior.

摘要

Recent advances in Cognitive Radio (CR) technology are reshaping modern wireless communications systems. Among numerous contributions CR technology has made, Radio Access Technology (RAT) multiplicity and Dynamic Spectrum Access (DSA) are of paramount importance. Advances in radio design now routinely allow multiple RATs to coexist on the same wireless device, further streamlining the design and implementation of DSA that provides flexibility in spectrum sharing. The flexibility enabled by the CR technology and Software Defined Radio (SDR) has even permeated up to the application layer where end users have been empowered to use wireless devices in many novel ways with smart phones and smart applications. In this dissertation, we have addressed two important aspects of CR networks, (i) resource allocation in multi-RAT enabled wireless networks; and (ii) impact and influence of end users' behaviors on the underlying protocol design. In the first part of this work, we study an example of the coexistence of multiple RATs devices in a network, namely a concept of Cognitive Digital Home (CDH). Motivated by the recent advances in radio design and wireless networking, along with the growth of multimedia home entertainment technologies, the concept of a cognitive digital home requiring spectrum coexistence of various devices and networks of networks is created. We have developed a framework for resource allocation in a CDH with a multiplicity of radio access technologies (RAT) such as cognitive radios and legacy radio devices supporting heterogeneous applications. We consider two channel access models in the CDH for addressing spectrum coexistence of legacy devices: (i) Pessimistic Controllability (PC) Model where the Home Genie node (HG) has no influence over legacy devices, and (ii) Switched RAT (SR) Model where the HG has perfect control of legacy devices. Distributed algorithms for maximizing sum rate and maximizing service capacity are designed using partial dual decomposition techniques. A distributed power control scheme is also designed for efficient use of energy. An admission control scheme based on pricing information obtained from the distributed algorithms is used to improve system feasibility. In the second part of this dissertation, we focus on the impact and influence of end users' behaviors on wireless systems and protocols by investigating the role of Prospect Theory (PT) in wireless network design. Prospect theory, a theory developed by Kahneman and Tversky, explains real-life decision making that often deviates from the behavior expected under expected utility theory (EUT). As a first step in exploring the role of PT in wireless networks, we consider a radio resource management problem where users follow PT and compare and contrast it to the case when users follow EUT. Specifically, we consider a random access game where selfish players adjust their transmission probabilities over a collision channel according to rewards received for successful transmission but also incur energy and delay costs. By analyzing the Nash Equilibrium (NE) achieved in a 2-player game, we prove under mild conditions that deviations from EUT of any player results in degradation of system throughput and increased delay and energy consumption. We also study N-player symmetric homogeneous games where all the users either follow only EUT or only PT, and observe similar results at the Nash Equilibrium. Finally, the framework introduced in the above random access model is extended to study an exemplary two-level data pricing model and compare and contrast service choices when users follow EUT and PT.