Improving the quality and reliability of traffic differentiation in IP networks

This article presents a modular scheduling architecture for multi-QoS metric differentiation in class-based IP networks. The rationale of the supported differentiation modules is presented, highlighting the distinct differentiation semantics that might be used to control the delay, loss and rate metrics associated with the traffic classes. The devised modules resort to several relative and hybrid differentiation models to bound QoS metrics on high priority classes. In the proposed scheduling architecture, the differentiation modules may act jointly in order to control simultaneously multiple QoS metrics. The results show that using simple and intuitive configuration procedures the proposed architecture is able to provide enhanced QoS differentiation behavior in IP networks according to the users and applications needs.