A study of Real-time fault detecting system using information filtering technique

规模和速度的扩展是互联网最重要的特征。建立一种有效的故障管理方法成为当前计算机网络的主要课题。在本研究中，我们提出了一种新的大规模计算机网络实时故障检测系统。首先构建了一个流量监控系统，用于采集网络流量中的错误数据包。然后，我们提出了一种新的算法——动态症状隔离算法（Dynamic Symptom Isolation algorithm）来有效地检测网络故障数据包，该算法采用过滤技术来抑制检测到的错误症状，从而彻底发现网络中存在于大量故障背后的小故障。实验表明，在大型网络环境下，新系统的故障症状检测性能是传统系统的两倍。

Expansion in scale and speed are the most important features of Internet. Establishing an effective way for fault management becomes the main topic for this computer networks. In this research, we proposed a new system for real-time fault detecting in large scale computer network. We first constructed a traffic monitoring system for collecting the error packets from network traffic flow. Then, we proposed a new algorithm, called Dynamic Symptom Isolation Algorithm, to detect the network fault packets in an effective way, in which filtering technique is applied to suppress the detected error symptom, so that the small fault existing behind the large number faults in the network could be found thoroughly. Our experiments showed that the new system's performance is as twice as the conventional system for detecting faults symptoms in large network environment..