Efficient End-to-End Transport of Soft Real-Time Applications

Abstract

Typically, real-time applications are well supported in environments where there is ample bandwidth available, enabling their transport within the stringent time constraints. Accordingly, non-realtime applications are well supported in environments where there is ample storage available resulting in low losses. For an emerging class of applications such as prerecorded video - referred to as soft real-time applications there may be an increased role that storage can play - in addition to bandwidth - in their effective transport. A scheme is proposed in this paper that takes advantage of available storage within the network and the end users, to efficiently transport prerecorded continuous media traffic across a number of network nodes.

The scheme is based on a scheduling policy implemented within the network nodes called Deadline Credit (DC) policy. Its objectives are (a) to utilise all (if possible) of the available capacity by allocating bandwidth dynamically to a number of data streams, (b) to distribute the available bandwidth fairly between all competing data streams and (c) to maximise the number of Application Data Units (ADUs) that are served by always giving priority to the one with the shortest time to expiration. Numerical results are presented and the induced ADUs loss rates under the proposed policy are compared against those of some alternative schemes.

This work is part of Zoe Antoniou’s Ph.D. research while with Northeastern University and has been supported by the National Science Foundation under Grant NCR 962811

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.