Richard Gibbens: research information
General information about research in the Computer Laboratory is available on the main website. Here I describe some aspects of my own research.
Here is a list of my publications.
The field of communciation networks presents many interesting problems and I have worked in a broad range of areas. The main attractions have been problems involving some form of distributed resource allocation often combined with a significant stochastic component.
Some of the main topics that I have worked on are as follows.
- Dynamic routing in telecommuncation networks
- Effective bandwidths for ATM networks
- Congestion pricing for packet networks
- Multipath routing in fixed and wireless networks
Road transport networks
I have been working on several aspects of road transport networks.
I have studied traffic flows, congestion and journey times on motorways specifically using loop detector data taken from sensors on, for example, the M25 London orbital motorway. In this work I have evaluated a simple but effective statistical methodology for short-term prediction of journey times which fuses together real-time data with data from a large historical repository.
The first figure shows a plot of speeds on the M25 over time and moving along the clockwise carriage way. The diagonal stripes are evidence of stop-and-go waves that propagate upstream during flow breakdown (that is, in the reverse direction to the motion of the vehicles).
The second figure shows the fundamental diagram relating speed and flow measurements taken minute by minute from a single loop detector. There are two clear regimes: a free flow regime and a congested regime.
As part of the TIME project within the Computer Laboratory I have been working with a post-doc, Andrei Bejan, on traffic on urban roads around the city of Cambridge. Here our source of data is a real-time feed of trajectories of GPS receivers fixed to buses. We have been able to derive some very interesting features of journey times and related them to time of travel and affects due to school terms. The fourth figure shows the daily profile of journey times for a given journey and uses quantile regression to clearly distinguish the affects of school terms.
This is an area of rapidly growing importance. There are many network problems and challenges embedded within the description of how energy networks might evolve in the coming years. I was able to learn much about this field and meet many of the active researchers through co-organizing the Energy Systems meeting held in Cambridge at the Isaac Newton Institute.
Some colleagues and co-authors
Prithwish Basu, Nigel Bean, Jean Bacon, Polina Bayvel, Andrei Bejan, Sid Chau, Jon Crowcroft, Tim Griffin, Jmaes Cruise, Saikat Guha, Phil Hunt, Frank Kelly, Peter Key, Robin Mason, Debasis Mitra, Erich Nahum, Fernando Ramos, Peter Reichl, Yunus Saatci, Richard Steinberg, Y-C Teh, Don Towsley, Stephen Turner, Phil Whiting, Stan Zachary, Ilze Ziedins.