Adaptive-weighted packet scheduling for premium service

This paper presents a new scheduling scheme to support a premium service in the differentiated service (DiffServ) architecture. It is based on weighted packet scheduling policies such as weighted round robin or fair queueing. The key feature of the new scheduling scheme is to change the scheduling weights of behavior aggregates adaptively. By adaptively adjusting the weights according to the dynamics of the average queue size of the premium service, the proposed scheme can achieve low loss rate, low delay and delay jitter for the premium service. Moreover, it requires neither rigid admission control nor accurate traffic conditioning to support the premium service in the DiffServ architecture. This adaptive packet scheduling is shown to absorb the transient burstiness of the expedited forwarding (EF) aggregate-which is caused by the traffic distortion inside the network-without incurring packet loss or increasing the queueing delay.