Research And Implementation Of Campus Network Fault Analsis System

With the development of information technology,people nowadays is getting more and more dependent on Internet in their life and work.Campus network is the basis of access to the Internet.It is of vital imporantce to locate the fault rapidly and accurately to guarantee the users'experience if campus network is in trouble.With the expansion of the network scale and the diversification of equipment,the fault relations in the campus network have become complicated.The traditional manual troubleshooting methods can't meet the actual demands.Alarm association rule mining technology can help to diagnose the faults by mining the historical alarm sequence to obtain the alarm association rules,and it has become a research hotspot in the intelligent fault location research area.Alarms in campus network have characteristic of timing correlation and space correlation at the same time.But existing alarm association rule mining algorithms relias solely on the timing correlation and ignored the space correlation,which leads to the problem of low correct rate of the mining result when used in campus network scenario.In order to solve this problem,and to fit the characteristic of campus network alert,this paper puts forward the Topo C-OPT(Topology Confidence-Optimizing)algorithm.On top of this,this paper designed and implemented the Campus Network Fault Analysis System.Firstly,based on the improvement of SDH-AARM(SDH-Alarm Association Rule Mining)algorithm,the Topo C-OPT algorithm is put forward for campus network.In the Topo C-OPT algorithm,the campus network topology is taken into account as one of mining factors.And then,topology correlation algorithm is constructed to calculate the correlation of alarm entities in the campus network topology.Furthermore,confidence optimizing model is built based on the above correlation.Experimental results show that in the campus network scenario,the proposed algorithm can improve the performance in terms of correct rate of mining result greatly compared with existing algorithm.With the proposed algorithm,the Campus Network Fault Analysis System is designed and implemented.The alarm data and the network topology data are obtained and normalized.And then,the system mines strong alarm association rules with Topo C-OPT algrithm.Furthermore,the system can analyze the fault of the real-time alarm sequence based on rules database.The system is deployed in a college campus network and tested.The results show that the system can mine strong alarm association rules with high correctness,and derivative alarms can be converged and faults can be analyzed correctly.