Abstract
In integrated services networks, the provision of Quality of Service (QoS) guarantees depends critically upon the scheduling algorithm employed at the network layer. In this work we review fundamental results on scheduling, and we focus on Packet Fair Queueing (PFQ) algorithms, which have been proposed for QoS wireline-wireless networking. The basic notion in PFQ is that the bandwidth allocated to a session is proportional to a positive weight φi. Because of the fixed weight assignment, the inherent in PFQ delay-bandwidth coupling imposes limitations on the range of QoS that can be supported. We develop PFQ with deterministic time-varying weight assignments, and we propose a low-overhead algorithm capable of supporting arbitrary piecewise linear service curves which achieve delay-bandwidth decoupling. Unlike existing service-curve based algorithms, our time-varying PFQ scheme does not exhibit the punishment phenomenon, and allows sessions to exploit the extra bandwidth in under-loaded networks.
Original language | English (US) |
---|---|
Pages (from-to) | 71-87 |
Number of pages | 17 |
Journal | Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology |
Volume | 30 |
Issue number | 1-3 |
DOIs | |
State | Published - Jan 2002 |
Bibliographical note
Funding Information:∗Work in this paper was supported by the NSF Wireless Initiative grant no. 99-79443. Parts of this work were presented at the Globe-com’2000 Conference. Original version was submitted to the Journal of VLSI Signal Processing on September 11, 2000. It was revised on June 6, 2001.
Keywords
- Generalized processor sharing
- Integrated networks
- Packet fair queueing
- QoS
- Service curves
- Wireless networks