Skip to main content

Analysis and Management of Wireless Networks with Selfish Users

Periodic Report Summary - WNET-GT (Analysis and Management of Wireless Networks with Selfish Users)

Project context and objectives

The emerging use of wireless technologies for data communication has brought novel challenges for the networking research community. The wireless medium incorporates novel system characteristics (power control, mobility). Due to the complexity of wireless networks, their performance can be largely affected by autonomous decision-making of the end users, not to mention the consequences of selfish user behaviour. When it comes to model and optimise the operation of modern wireless networks shared by self-interested parties, game theory appears to be a natural tool. The general objective of this project can therefore be summarised as follows: Identify game-theoretic situations arising in wireless networks, analyse them, and propose protocols and low-complexity management schemes for proper network operation. The main objective in designing such network protocols is to allow network users to obtain their required Quality of Service (QoS) in a distributed way, while maintaining network-wide efficiency and fairness among the users.

The first two years of the project were carried out with full accordance with the project objectives. In particular:
- (i) Several new scenarios which arise in wireless networks have been carefully examined;
- (ii) Distributed MAC (media access control) protocols that support QoS have been designed for these scenarios;
- (iii) Efficient management schemes for random access wireless networks have been suggested and analysed.

These developments were carried out by closely following the methodology which has been suggested in Annex I. Specifically, we have extended the "basic model" of a wireless uplink control problem (described in Section B1.2) in several aspects:
-(i) more complicated assumptions on the physical channel;
-(ii) additional user utilities;
-(iii) considering general network topologies and different kinds of interactions between mobile users.

Our results show that wireless protocols, which allow end-users to autonomously adjust certain parameters of their protocol, can achieve adequate performance, as long as the degree of freedom is carefully chosen. Going beyond the 'basic model', I have begun studying the user decision process in cognitive-radio networks, an emerging domain within wireless-network research. In particular, the question I have focused on is how a spectrum owner can maximise its profits, in view of the alternative that Secondary Users have in the form of free "white-spaces" spectrum. My research results have been extensively published in leading journals and conferences. Moreover, I am in the process of writing a monograph on network games, part of which will be dedicated to my present research on wireless networks.

Alongside the scientific results, I have broadened the scope of my research in terms of methodology and tools. Motivated by wireless network applications, I have developed a novel game-theoretic framework for the analysis of general non-cooperative games. The main idea of the framework is to study the static and dynamic properties of games by their relation to a "close" potential game. Such close potential-game can be extracted through our decomposition theorem, which relies on a novel flow representation for finite games in strategic form. In addition to the above game-theoretic machinery, I have extended the set of tools which I employ in my research, including online algorithms, queuing theory and network pricing.

My plan for the third year is to continue studying wireless networks shared by self-interested users, with emphasis on new architectures, such as cognitive radio networks. At the same time, I intend to dedicate substantial efforts to continue developing game-theoretic tools that will enhance the predictability and controllability of complex wireless domains. By such combination of theory and application, I truly believe that this project will have a substantial impact on understanding how current wireless networks perform, and on how future wireless networks should be designed.