Network Fault Diagnosis Research Based On Immune Agents

With the rapid development of computer network technology, thesize of the network is becoming larger and larger, and the sum of peoplewho use the network is growing dramatically, the current networkmanagement and assert is more and more difficult to adapt to the user’srequest. The causes of the network faults are varied, and they alwayscontain a lot of information. So it is very hard to solve the entire networkfaults by the tools and the methods which are provided now.Therefore, this article introducing the mobile Agent technology inthe network fault diagnosis problem, through the artificial immuneprinciple complete detector training, network fault diagnosis model is putforward, the model with the basic structure of the Agent as the frameadopts step by step, the diagnosis of thought, and using the Agent in eachlevel step diagnosis strategy alliance. Then based on clone selectiontheory complete detector training immune algorithm, this algorithmfocuses on detector clone selection of group and the introduction ofdynamic optimization parameters, avoid to produce the phenomenon ofpremature convergence and local optimum solution, and introduces theclassification of detector of ideas, to a wide range of fault types areclassified, shorten the diagnostic time. Finally adopts the Aglet of IBM’sdevelopment platform and the Java programming languageimplementation, design the function of each module, by collectingnetwork interface set of failure data for simulation, verify the article putforward the feasibility of the algorithm. At the same time, by studying theunknown fault types, further proves the effectiveness of the proposedalgorithm.In this paper, research work is the core of the design of network faultdiagnosis system based on immune agent and its application in faultdiagnosis. First introduces the concepts related to biological immunesystem and agent technology and working principle. Then for sometypical intelligent network fault diagnosis methods are analyzed incomparison. Finally on the basis of these work is put forward to the basic structure of the Agent as a framework of network fault diagnosis model,and use the improved clone selection algorithm to complete training ofunknown fault detector. The immune model is studied through simulationexperiment, the performance and the effectiveness of network faultdiagnosis.