Premium
Proportional rate‐based congestion control under long propagation delay
Author(s) -
Cavendish Dirceu,
Oie Yuji,
Murata Masayuki,
Miyahara Hideo
Publication year - 1995
Publication title -
international journal of communication systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.344
H-Index - 49
eISSN - 1099-1131
pISSN - 1074-5351
DOI - 10.1002/dac.4500080202
Subject(s) - computer science , network congestion , network traffic control , network packet , computer network , transmission (telecommunications) , packet loss , telecommunications
The occurrence of congestion severely degrades the performance, especially in high‐speed networks with long propagation delay relative to packet transmission time. This is because it is very likely that a large number of transit packets are flying over such networks at a time and many of them will get lost if they arrive at some congested node. Among ways of controlling traffic to avoid congestion, this paper deals with a class of rate‐based congestion controls (RBCC) which regulate input traffic based on feedback information about network congestion status. Although such mechanisms have been studied recently, the mechanism to be considered, referred to as proportional RBCC, differs from them in that the amount of input traffic allowed to enter in the network is proportional to the network resources available, keeping a closer track of the network resources than the previous ones, where the network status is described by two states, congested or not. The mechanism seeks an optimal network operating point, adapting to dynamic changes of carried traffic. We carry out the stability analysis and the transient analysis of some performance measures related to the proportional RBCC, and also address fairness aspects in providing control to various source‐destination connections, while K. W. Fendick et al. have analysed the stability of a general class of RBCC. Furthermore, by making a comparison between the control mechanism analysed here and the ones based on binary feedback information, we show the effectiveness of the former.