A predictive self-tuning fuzzy-logic feedback rate controller
IEEE/ACM Transactions on Networking (TON)
Journal of High Speed Networks
Design of Robust Congestion Controllers for ATM Networks
INFOCOM '97 Proceedings of the INFOCOM '97. Sixteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Driving the Information Revolution
The Effect of Bottleneck Service Rate Variations on the Performance of the ABR Flow Control
INFOCOM '97 Proceedings of the INFOCOM '97. Sixteenth Annual Joint Conference of the IEEE Computer and Communications Societies. Driving the Information Revolution
Prevention of deadlocks and livelocks in lossless backpressured packet networks
IEEE/ACM Transactions on Networking (TON)
A new single-bit feedback congestion scheme for ATM networks
Computer Communications
Evaluation of a start-stop protocol for best-effort traffic in ATM networks
Computer Communications
Hi-index | 0.07 |
This paper presents a new feedback congestion control mechanism for the flow control of the best-effort available bit rate (ABR) traffic in ATM networks. This new mechanism belongs to the class of feedback control schemes that ensure no data losses and operate based on simple `stop' and `start' signals. A novelty presented by this paper is a methodology which, for a given set of desired properties, leads to the specification of the corresponding control algorithm. For the case of a single connection, the algorithm can operate with the theoretically minimum possible buffer size. Interestingly, the algorithm obtained has a different structure than the previous schemes; it does not operate based on fixed high and low thresholds. A new congestion control mechanism is subsequently derived for the flow control of multiple connections. The new scheme is exercised hop-by-hop and on a per-connection basis. This scheme allows connections to share memory and bandwidth resources efficiently within the network. The performance of the new scheme is also presented, and its statistical multiplexing efficiency is demonstrated. The measures investigated include buffer occupancy, average delay, overhead due to the protocol signals, and sustained throughput. In the case of long propagation delays, the buffer savings achieved by the new scheme are substantial