QoS-enhanced TNPOSS Network Model and its E2E Delay Bound for Multimedia Flows

In our previous work, we proposed the TNPOSS (To Next-hop Port Sequence Switch) network which can achieve scalable fast forwarding and is suitable for delivering multimedia flows. However, TNPOSS network has no Quality of Service (QoS) tools to provide better QoS guarantee for multimedia flows which often have long-range dependence (LRD) property. To enhance the QoS ability of TNPOSS network, this paper proposes a QoS-enhanced TNPOSS (QTNPOSS) network model by introducing Fractal Leak Bucket (FLB) shaper and Weighted Fair Queuing (WFQ) scheduler into the original TNPOSS network. In the new QTNPOSS network, each packet of multimedia flows is shaped by the FLB when arriving at a node and is scheduled by the WFQ before being outputted. Based on study above, we further analyze the end-to-end (E2E) delay bound of QTNPOSS network by means of the network calculus theory which is an effective mathematical tool on analyzing the worst-case QoS performances of networks. The service curve and the formulation of E2E delay bound of QTNPOSS network are presented and proved. Extensive numerical experiments show that both the LRD property of multimedia flows and the WFQ weight have influences on the E2E delay bound, and the WFQ weight influences the E2E delay bound more greatly than the LRD property does.