Pricing the Internet

Other Resources

Summary

This page lists some background resource material on shadow pricing. After having seen our views on how the Internet should be charged for, you can see what other people think.

Background material on Shadow Pricing

Gibbens and Kelly Resource pricing and the evolution of congestion control (paper)
Gibbens Resource pricing and the evolution of congestion control (web site)
Kelly, Maulloo and Tan (1998) Rate control in communication networks: shadow prices, proportional fairness and stability (paper)
Tan (1998) Rate control and user behaviour in communication networks (paper)
Microsoft Research, Cambridge Congestion Pricing and a Distributed Game (web site)
Key and McAuley (1999) Differential QoS and Pricing in Networks: where flow-control meets game theory (paper)
Crowcroft and Oechslin (1998) Differentiated end to end internet services using a weighted proportionally fair sharing TCP (paper)
Wischik Pricing the Internet (talk)

Other ideas about resource pricing

Floyd and Fall (1998) Promoting the use of end-to-end congestion control in the Internet
Feng, Kandlur, Saha and Shin (1999) Blue: A New Class of Active Queue Management Algorithms
Gupta, Stahl and Whinston (1995) Priority pricing of integrated services networks
Clark (1995) Adding service discrimination to the Internet
MacKie-Mason and Varian (1993) Pricing the Internet

People who would pay more for a better service cannot do so
I will venture a total speculation that a simple scheme with only two levels will provide an effective allocation of service
New mechanisms and pricing must go hand in hand to provide a range of service levels
it should be implemented in such a way that users have incentives to make the right decisions
The point of a fair allocation is to assure the users that they are being treated equitably. The problem is to find a useful definition of fairness
The end-to-end congestion control mechanisms of TCP have been a critical factor in the robustness of the Internet
An increasing deployment of traffic lacking end-to-end congestion control could lead to congestion collapse in the Internet. The network must now participate in controlling resource utilization.
To control the overall sending rate, one must control the congestion feedback received at the source of the data
High-bandwidth flows should be regulated by having their bandwidth use restricted at the router
Experience using the network provides a pragmatic sense of what the response will be to service requests of various sizes at various time of day. This idea of expectation as opposed to guarantee is an important distinction
drastic unfairness that results from TCP flows competing with unresponsive UDP flows
The TCP protocol itself is subject to change

Damon Wischik D.J.Wischik@statslab.cam.ac.uk