Abstract

Scarcity of frequencies combined with the demand for more bandwidth is likely to increase the need for devices that provide high wireless bandwidth in limited areas while using a wired network to carry data over longer distances. Examples of such devices are Wi-Fi routers and femtocells and future devices that use TV whitespace. To utilize the available frequencies more efficiently radios must be able to find other users of the frequency bands and adapt so that they are not interfered. As transmitters hundreds of kilometers away may cause as much interference as a transmitter located next door, this mechanism can not be based on location alone. Central databases can be used for this purpose, but with thousands or millions of radio devices to coordinate a centralized system may not always be ideal. In this talk, we describe a decentralized protocol and architecture for discovering interfering radio devices over the Internet. The protocol has low bandwidth-, memory- and processing requirements, making it suitable for platforms with limited resources. We evaluate the protocol through simulation in network topologies with up to 1 000 000 nodes, including topologies generated from three municipalities in Norway.

This talk is based on joint work with Brage Ellingsæter, Torleiv Maseng and Jon Crowcroft.

Bio: Magnus Skjegstad is a PhD student at University of Oslo and Norwegian Defence Research Establishment (FFI). He is currently visiting the NetOS group at Cambridge University.