A fair queueing algorithm for multiple-streams delay-bounded services

In this paper, we propose a resource reservation and scheduling algorithm which guarantees the end-to-end delay constraints for the processing of single-source multiple-stream delivery. User applications initiate the requests by specifying the tolerable delay and priorities reflecting the preference for each stream. In order to provide efficient flow scheduling subject to the fairness and delay constraints, we consider a mechanism in which the resource reservation and scheduling are integrated into a framework so that the queueing delay can be taken into the consideration at the reservation phase. Packets are scheduled in a packet-by-packet base, the fairness characteristics in each scheduler is approximated by emulating the max-min fair discipline in the corresponding GPS server. Instead of tracking the fair utilization before the packet can be served in WFQ, the bandwidth share is monitored after the packet is sent. This approach can significantly reduce the computational complexity resulting from WFQ while maintaining the long-term fairness. The end-to-end delay is derived from the latency in each link and the worst-case waiting time at each scheduler. Examples are illustrated to show the performance, the results lead to high satisfaction in terms of delay and fairness.