Research On Distributed Network Performance Monitoring Based On Mobile Multi-Agent

The ever-increasing scale and complexity of modern networks brings out new requirements and challenges for network management. Network performance mamagement,as one of the basic functions of network management,should adapt to the development of networks too.A network performance monitoring system based on mobile multi-agent has been proposed in this thesis.It makes use of agent's mobility and cooperation and combine advantages of hierarchical management model and cooperative management model to realize distributed and efficient network performance monitoring.This thesis introduces basic concepts of network management,summarizes agent technologies,surveys those existing network management technigues based on mobile agent and multi-agent, analyses their advantages and diavantages,and brings forward the mobile multi-agent based performance monitoring framework.The design methods of the components in the framework are also provided.The proposed system includes three main algorithms:(1)Network topology discovery algorithm based on mobile multi-agent.After summarizes the logical topology discovery algorithm based on route table,and researchs on the shortcomings of classic physical topology discovery algorithm based on address forwarding table,a physical topology discovery algorithm based on STP has been proposed to achieve accurate and efficient discovery of connections between switchs .On the basis of these algorithms realization algorithm based on mobile multi-agent are brought forward.(2)Monitoring agents deployment algorithm. One of the critical problems in distributed performance monitoring is how to place proper number of monitoring agents in proper locations in a managed network can improve monitoring efficiency, that is,decrease monitoring traffic and response time.The proposed algorithm is a distributed location computation method,in which the number and locations of child agents are computed by their parent agents with local route information.Simulation results indicate this distributed agent system can achieve proximate optimization and reduce monitoring traffic comparing with centralized counterpart.(3)Monitoring agents self-adaption algorithm.The algorithm is triggered when network topology changes,or nodes and links failures occur, by adjusting some agents' deployment,it can draw the whole monitoring system back to proximate optimization state .This algorithm add flexibility and scalability to the proposed monitoring system.