Congestion Pricing for Congestion Avoidance
Peter B. Key, Derek McAuley, Paul Barham, Koenraad Laevens · 1999
Abstract This paper describes the use of Congestion Pricing as a means of providing Congestion Control and Differentiated Quality of Service. The application of the proposed technique to the Internet Protocol has the advantage that it can be simply implemented using Explicit Congestion Notification. In particular: the network mechanism is independent of higher level protocols; the end systems can continue to exhibit current TCP behaviours; new multiprotocol flexibility is made available to end systems and users. Architectural issues are discussed, including important aspects of aggregation and charging. We describe our methodology for assessing the scheme via a distributed network simulator. Initial results are presented which compare and contrast various adaptive strategies that achieve a variable range of TCP-like behaviours. 1 Introduction The current control mechanisms of the Internet are not well suited to offering different Quality of Service (QoS) to different channels. TCP, the dominant transport control protocol, is appropriate for "besteffort " traffic, and aimed at data where lost information has to be retransmitted. However, other types of traffic, such as streamed video or audio, are sensitive to delay as well as packet loss. Such traffic is often carried over UDP, which has no generic feedback signals or error recovery. Certain commercial IP offerings make use of the TOS field to give a two-level service offering, though this is only possible when all routers are under the provider's control. The Diff-Serv proposals to provide a small number of QoS classes can be seen as an extension of this idea. Within the Diff-Serv community, Explicit Congestion Notification (ECN) has been proposed as means of triggering active queue management disciplines to implement congestion avoidance.