In this project the PI plans to investigate a game-theoretic analysis of network flows with heterogeneous delays, the principal application being to analyze the impact of prioritization of traffic on the internet and consequently, to shed light on the debate on net neutrality. With the internet becoming an essential aspect of communication and big data transfers becoming more frequent, Quality of Service (QoS) becomes an important aspect of network design and usability. The consequent network traffic prioritization has given rise to the debate on net neutrality and the PI will investigate the impact on the efficiency of networks as a consequence of negating this principle, given the rational self-interests of the users. While the inefficiency of selfish routing has been studied in the past, this project will extend the state of the art by utilizing heterogeneous delay models that reflect traffic prioritization as well as bandwidth reservations. Economic aspects will also be considered. The results will also have applications in the field of transportation networks.
The project aims to model the routing of multiple types of traffic arising from independent autonomous agents on a network via multi-commodity network congestion games with heterogeneous delay functions. Heterogeneous delay functions, such as polynomial delays and the generalized processor sharing (GPS) queuing function, model the impact of differentiated or heterogeneous traffic such that prioritizing a commodity increases its impact on the delays of all the commodities on the link. Other applicable delay functions and consequent results from queuing theory will also be applied. The project will study the existence of Nash equilibrium and properties such as the price of anarchy in the context of both heterogeneous traffic as well as service classifications that will induce bandwidth partitions. Equilibrium pricing in network markets will also be considered.
Results obtained in this project will further current research in the theory of congestion games and may shed light on the debate over net neutrality. Educational impact of the project will be via the training of graduate students in algorithmic game theory and development of related courses.
