Abstract
Myths about congestion control are examined, and an explanation of why the trend toward cheaper memory, higher-speed links, and higher-speed processors has intensified the need to solve the congestion problem is provided. A number of proposed solutions are described, and a classification of congestion problems as well as their solutions is presented. The reasons why the problem is so difficult are identified, and the protocol design decisions that affect the design of a congestion control scheme are discussed. The author's recent proposals, namely, timeout-based congestion control, a DECbit scheme and a delay-based scheme for congestion avoidance are described, and areas for future research are suggested.
All Related Versions

This publication has 24 references indexed in Scilit: