Network Fault Location And Detection Technology

With the rapid development of information technology, computer network has become an essential infrastructure in people's life. At the same time, size and complexity of the network increase explosively. This situation makes the traditional way of network fault diagnosis relying on manual work in trouble. People need the intelligent network fault diagnosis technology urgently. This paper researches two important issues about network fault diagnosis: fault location and fault detection.Because of the connectivity of network, there are some correlations between network faults. Single network fault may cause lots of phenomenon. And every phenomenon will be captured by network fault diagnosis system as an isolated event. Network fault location should find the source fault from several fault events. In this paper, a fault location algorithm based on correlation graph and case base is proposed. This algorithm can process the complicated correlation between multi-faults, and improves the location efficiency by case base. Experiment demonstrates that the algorithm has good effect.Fault detection is responsible for establishing mapping relationship between fault characters and reasons. And when there have a fault, find the possible reason by analyzing the real-time fault character parameters. In this paper, we research several kinds of existing techniques on fault detection. To solve the disadvantages of the Fault detection using BP Neural Network, such as large size and slow training, we reduce the training set through Rough Set approach. And we choose a heuristic algorithm to ensure the effectiveness and speed of the reduction. Simulated experiment demonstrates that the improved method reduces the size of the neural network and improves accuracy rate of the detection.