The Algorithms Research On QoS Anycast Route

With the development of Internet, Network service requirements exceed the capabilities of the network, which causes a great influence to application of Quality of Service (QoS). In order to enhance the service ability and improve network flow distribution, we usually duplicate servers in the network. Anycast service is a new service model supporting distributed duplication. More and more applications are related to anycast, so anycast is a standard service model defined by IPv6. This thesis concentrates on the study of QoS anycast routing algorithms and related technologies.The thesis starts with an analysis on the general model in QoS anycast service and basic QoS anycast routing algorithms. With the standpoint of load-balanced, this thesis presents a QoS anycast routing algorithm based on network link idleness probability. This algorithm combines the network total loads with request bandwidth, and searches for a QoS anycast routing path. The experimental results show that the new algorithm can balance loads of network and server efficiently, and has higher acceptance rate of requests. Because the network state information is not always accurate, this thesis brings forward a QoS anycast routing algorithm based on safety rate. Utilizing inaccurate network information and considering update policy, this algorithm gives a safety function to every link in network, and hunts for an anycast routing path through it. The experimental results show that the proposed algorithm has good load-balanced with highly acceptance rate under different network parameters. Because traditional QoS routing algorithms ignore the trade-off and total performances, this thesis proposes a QoS randomized routing algorithm. This algorithm uses the network average idleness rate as an evaluation parameter, and searches for a transmission path according to a presented safety formula. The experimental results show that the new algorithm can improve cability of network loads and optimize performance of the whole network. According to ideology of designing a network simulator, we design an event-based simulator. The performances of all algorithms in the thesis are evaluated through the simulator.