Game Theory Application to Interdomain Routing

Abstract

This thesis proposes a game theoretic analysis of interdomain routing. In the two following chapters we try to capture some of the intricacies of the current Internet routing protocol, the Border Gateway Protocol (BGP). The first chapter of the thesis paper presents a survey of recent adv ances in the application of game theory to model the behaviour of interdomain routing. In these models, the participants of the interdomain routing game are represented as strategic agents seeking to improve their benefits through the manipulation of the interdomain routing protoco l BGP. The main results achieved over the last few years in this field include models to an alyze the stability of the interdomain routing and the design of mechanisms that guarantee BGP to be incentive-compatible with or without monetary transfers. However, over the last years, the research community has been deeply concerned about the scalability issues that the Internet routing is facing. As the Internet popularity grows, so do the network resources needed in order to sustain its worldwide availability. In the second chapter of this thesis we consider a commons model in which the Global Routing Table (GRT) is a public resource. We use this model to study the economic incentives the ASes have for deaggregating their assigned address blocks. We evaluate the efficiency of the global routing system, the properties of the game equilibria and we examine its relation to the social welfare point of the considered game setup. We find that the str ategy adopted by the ASes in the interdomain is not an overall optimum strategy and it leads to an inefficient exploitation of the common resource. Therefore, we prove that the GRT, just like any common natural resource, “remorselessly generates tragedy”, following Hardin’s game theoretic analysis on the tragedy of the commons. Finally, we introduce in the model a pricing mechanism that aims to avoid the tragedy of the Internet routing commons.