QoS Routing Based On Efficient Utilization Of Network Resources

Future network would provide various kinds of services, such as net meeting, synchronous transmission of data and video in distance medical systems. All these services are in demand for the guarantee of the quality of services. In the architecture of QoS, QoS routing is to select routes for the traffic based on the QoS requirements and available network resources. On the one hand, QoS routing should ensure that the selected route satisfies the quality requirement of traffic. On the other hand, it should improve the performance of network to maintain the transmission of admitted traffic flows.This dissertation undertakes a detailed investigation of the relationship between QoS routing and the stability of network nodes' state. Unbalanced network resources assignment will bring up congestion thus make change of network node state. This change will result in a failure for QoS routing which will exacerbate the performance of network. Therefore, we study how well a QoS routing can utilize the network resource and to ensure the stability of node state by rationally assignment of network resources.We first propose a path computing algorithm combining with estimation of network cost. We develop a new cost function which takes into account both the current utilization of network and future impact of the traffic to the network as a whole. Using this metric, we select routes for traffic that brings the least influence to the network. Besides, we introduce a preliminary admission control algorithm. Whether a kind of traffic should be admitted not only depends on whether the selected route can satisfy the traffic's delay requirement, but also depends on whether the available network resources during the traffic's duration can accommodate the traffic. This admission control scheme takes the time characteristic of network resources, which can be used to optimize the performance of network.We construct a network simulation platform based on a popular network simulation software. This platform includes modules of topology, traffic source and routing. It can perform different routing, admission control and scheduling algorithms and can also analyze the impacts of different configuration parameters on the routing process. Using this platform, we compare the performance of several classical routing algorithms under two famous topologies. The results show a improvement in the utilization rate of network links and in the stability of network nodes state.