One Approach Towards Fair Anycast Routing Forward Algorithm Based On Game Theory

In the research project of Important National Basic Research Program of China (973 Program), existentable massive information system software design theory, the network existentablity is one research subject. With the born of the NGI, the request of network application is being much more than before and the network robust property now is contained in the network existentablity. It is one subject of this project to minimize the average delay time during the routers fowarding data packets.For fault-tolerance and fair load distribution purpose, many mordern internet applications may require that a group of replicated servers dispersed widely over the world. The anycast is a new communication style defined in Ipv6 provides the capability to route packets to the nearest server. Better quality of service(QoS) can be achieved by this kind of computing paradigm. However, before anycast can be realized , more research need to be done, includes anycast route mechanism.The load distributions on the routers output paths influence each other. Which policy is used by this route path and how much load will send to this route path will cause the other paths to take different policy and get how much data load. Therefore, it is necessary to use game theory to discuss their load distribution.With related game thoery konwledge, this article provides one method MRS to calculate the data distribution proportion of all the router's distribution paths. It is used in the WRS route method. Meantime, it provides the algorithm of MRS in this article. MRS has taken account of the minimization of the average delay and the fair load distribution. With the examination, it is proved that this method has fairness property. At the same time, this article give the comparison of average dalay time between this method and other two, Proportional and Overall Optimal. The result is: when the load is under 50% of the sumation of all the route path capability, its average delay time is close to that of overall optimal; when the load is between 50% and 80% of the sumation of all the route path capability, its average delay time is between the other two methods; when the load is above 80% of the sumation of all the route path capability, its average delay time is close to that of proportional.