Research Of Network Routing Optimization Based On Intelligent Agent

In the network, the currently used Dijkstra shortest path algorithm and routing protocol and others are based on the network topology and configuration parameters to select path without considering the network load, link performance and router current flow. They may transmit data with a few links or nodes. Large amount of data transmission through a few links or nodes is easy to cause the network congestion and bottlenecks, and increase packet delay and packet loss rate. What's more, it may cause network paralysis. Besides, unbalance of data load reduces utilization of resources and equipment and shortens the service life of the equipment with heavy load nodes. So, how to improve router transmission and load equilibrium strategy, how to enhance the network transmission efficiency and throughput and then reduce the network congestion and delay and how to properly use network resources are Internet's challenges today.Firstly, this paper introduces source routing technology and intermediate routing technology. And it constructs the strategy of network layers so as to improve the network scalability and give the full play of the source route network reliability with the intermediate routing technology. In the source route, the source node contains the shortest path set of all pairs of nodes in the network. It will specify transmission path, while intermediate nodes on the path needn't to reserve routing information. At the same time, source route is a multi-path route, which can improve the network communication quality by specifying the separation paths between the source and end node. However, source route does not apply to large-scale network with high cost and poor scalability. And in the intermediate routing technology, the messages are send from source nodes to a designated intermediate routers, and then from the router to destination nodes, rather than from source nodes to destination nodes directly. Therefore, we can combine the source routing technology with the intermediate routing technology to achieve complementary advantages by setting specific intermediate routing nodes in the network.Secondly, the node importance and the network performance after establishment of intermediate nodes was analyzed, especially the probability of the network path failure when introducing into a single intermediate routing node and several intermediate network nodes respectively.Thirdly, intelligent agent technology was introduced and applied to the load information collection, update and load distribution to get a new load balancing algorithm. Broadcast or polling method are used to extract load information in most of load balancing system, which lacks of flexibility and takes up a lot of network resources. Intelligent agent has sorts of advantages, such as initiative, self-learning ability, reasoning, mobility, collaboration and so on. Network overhead can be reduced as well as network reliability and robustness can be improved, if we applied intelligent agent to the network communication. Intelligent agent was placed in the middle of routing node, which responsible for collecting and updating the load information of nodes and links. And intelligent agent was placed in the general node, according to the load information to calculate weight of the load path. The weight of the load path can be used to distribute the load with the way of weighted distribution, which is more effective in achieving rational use of resources, increasing network throughput and reducing network latency and packet loss rate.Finally, the simulation software was developed according to C # in .Net framework 2.0 platform. And then, the improvement of network performance after the introduction of intermediate nodes and load balancing algorithm was simulated and validated. The experiment results show that the rate of not up to the path of the network after the introduction of intermediate routing nodes reduce greatly and improve the network congestion situation effectively. In addition, the load equalization algorithm which integrated the intermediate routing nodes and intelligent agent technology, greatly improve network communication performance as well as reduce the packet loss rate, especially when a large traffic, the improvement becomes more apparent.