Research And Application Of Ant Algorithm Based On Core Router

With the widely application in internet, its scale has become larger and larger, and its flow increased quickly too, it is forced to develop its transmission to high bandwidth. So the construction of network with high speed and wide bandwidth becomes essential.The reasonability and high efficiency in routing can not only guarantees the net works well, but also can increase the rate of connection. In this way, it can avoid collapsing because of unbearable heavy burden in swiches but also can decrease the cost of the network. Increasing the rate of connection in network largely depend on the methods of routing. Therefore, the dynamic routing method in TCP/IP network is becoming more and more important. The ant algorithm can choose a superior path availably, but another problem that neglected in actual network is that, once the superior path is formed, all dates will be delivered from this path, in this case, the router on this path, especially the router in center node of the network (several paths hands over to remit a place) will bear the enormous data delivering. As a result, the "bottle-neck" phenomenon formed easily. Currently, a method to solve this problem is to establishes a core router in center node of the network, which has the commutation capacity attains or over thousand MBits and has super-speed port. While it is larger increasing the network costs, and can't resolve the data jam of non-core router in superior path.In this case, this paper presents an improvement ant algorithm which based on the core routers: on each core router, sending ants to look for the superior path mutually between each other, it will be much quicker than the traditional ant algorithm which is looking for the superior path by pass the whole network. On the other hand, in the superior path, the algorithm se two values on non-core routers to avoid jam of data when the non-core router A is becoming jam, the near core-router treat A as a " stumbling block", so it will choose another second superior path which is not including A, in this way, the data is share by the second superior path. When the value reduced, it will choose the first superior path again. As a result, this algorithm can not only find the superior path, but also avoid jam of flow.