RODYNE: Robust and Adaptive Routing in Dynamic Networks


The aim of this project is the development and rigorous analysis of fundamental methods for efficient and robust communication in dynamic networks.

In most general terms, dynamic networks are networks of dynamically changing network characteristics. For example, nodes or communication links may enter or leave the system, or the capacity of nodes or communication links may change. There are many scenarios in which dynamic networks occur. For example, a fixed interconnection network might be under physical or electronic (such as denial of service) attack causing communication links to go down, or we may have a wireless network formed by users that move around. Another increasingly popular scenario is communication via overlay networks such as peer-to-peer networks, where there is a continuous stream of users entering or leaving the network. Also, the Internet can be seen as a dynamic network, since TCP connections or communication links might experience large differences in their availability over time, offering a low bandwidth at one time and a high bandwidth at another time.

Why is it interesting to study communication in dynamic networks? The development and deployment of powerful and flexible communication systems that are accessible by large parts of the population is seen as a key component for the transition to an information society. While this has been dominated in the last 10 years by the establishment of wired connections to the Internet, this is now shifting more and more to wireless connections, since wireless connections offer a much higher mobility and flexibility. Many companies and universities have already deployed wireless access points that allow employees and/or students to access their local area network or the Internet via mobile devices such as laptops and palmtops. So far, wireless communication has mostly been restricted to small areas, but several projects are already under way to establish large-scale wireless networks accessible by large parts of society.

In fact, Wi-Fi Internet access is slowly being spread to cover more and more area in cities with high concentrations of techies, such as San Francisco, Seattle, and New York. These enthusiasts have formed groups -- such as SF Wireless and the Bay Area Wireless Users Group (BAWUG) in the San Francisco area -- that are working together to build free high-speed wireless networks. Also, an increasing number of hotels, airports, and even city governments are touting free Wi-Fi access for their patrons and residents. Moreover, Earthlink and Atlas Broadband have recently made deals with wireless network integrators Boingo and Joltage to create widespread Wi-Fi networks that users can log onto for a fee. Also, Boeing is currently funding research in large-scale ad-hoc networks, and even car companies are interested in wireless technology. For example, in a project with the code name Moteran, Mitsubishi, Toyota and VW tried to develop devices that would allow future cars to interconnect to autonomous wide area networks.

In parallel to these activities, overlay networks formed by users connected to the Internet, also known as peer-to-peer networks, have become extremely popular in recent years. While this was initially motivated to escape censorship and law suits (consider, for example, the case of Napster), this technology has also become highly interesting for businesses, because in addition to being good for swapping soundtracks, peer-to-peer networking is a way to harness the dormant processing power in desktop PCs. The potential benefits are so great, in fact, that global players such as Hewlett-Packard, IBM, and Intel want to standardize and commercialize the technology. For this purpose, they launched together with several start-ups the Peer-to-Peer Working Group. To obtain a critical mass of industry support, this consortium recently decided to join the Global Grid Forum.

Due to these reasons, communication in wireless networks and peer-to-peer networks is currently a rapidly evolving, highly active research field. Fundamental research will be important to identify efficient and robust communication strategies for these types of networks - which represents the research goal of this project - and people are needed with a solid background in theory and practice of communicating in dynamic networks to lead research and development in this area in the future - which represents the educational goal of this project.

Faculty members:

PhD students:

Other students:



Christian Scheideler
Last modified: Tue May 20 2003