Abstract
Purchase PDF (1322 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (1)
Purchase PDF (500 K)
doi:10.1016/j.comnet.2004.11.006 
Copyright © 2004 Elsevier B.V. All rights reserved.
P. Balaouras
, 
, , and I. Stavrakakis
Received 4 August 2003;
References and further reading may be available for this article. To view references and further reading you must purchase this article.
Abstract
A novel feedback-based rate adaptation scheme is introduced and investigated in this paper. Its main innovative characteristic is the modulation of the rate increment by the distance between a flow’s present rate and an assumed targeted maximum rate as dictated by the associated application. The previous along with the shaping of the rate decrement by the reported flow’s losses are responsible for a dynamic and self-adjusting behavior that is shown to improve convergence to fairness, the oscillatory behavior of the rate and the induced packet losses when compared with the basic Additive Increase Multiplicative Decrease (AI/MD) scheme. Numerical results illustrate the good properties and intrinsic advantages of the proposed scheme both under the considered modeling assumptions, as well as under more real networking conditions by employing the ns-2 simulator. A brief comparison of the proposed scheme with the TCP-compatible schemes TFRC, IIAD and the non-AI/MD schemes AIPD, LIMPD, is included as well. Because of the aforementioned induced behavior and assumed flow’s characteristics (min and max rates), the proposed congestion control scheme seems to be appropriate for regulating the rate of streaming applications.
Keywords: Elastic flows; Rate adaptation schemes; Congestion control; Multimedia; Smoothness; AI/MD
Article Outline
- 1. Introduction
- 2. Description of the distance weighted additive increase and loss rate dependent multiplicative decrease scheme
- 3. Convergence to efficiency and fairness
- 4. Analysis of the behavior of the DWAI/LDMD scheme
- 4.1. Responsiveness to BW availability of the DWAI policy
- 4.2. Periodic oscillatory behavior of the DWAI/LDMD scheme
- 4.3. Size of oscillations and packet loss
- 4.4. Throughput
- 5. Discussion on the comparative performance of the DWAI/LDMD and AI/MD schemes
- 5.1. Convergence speed to fairness
- 5.2. Convergence to efficiency
- 5.3. Smoothness and packet loss rates
- 6. Simulation results
- 6.1. Evaluation of the DWAI/LDMD scheme
- 6.2. Evaluation of the DWAI/LDMD scheme in realistic environments
- 7. Related work
- 8. Conclusions
- References
- Vitae
