The increasing complexity of modern network has motivated the development of different fault detection and isolation approaches for the purpose of supervision. Currently, packet monitoring has become a standard technique in network fault detection, but when applied to a large-scale network it yields a high volume of packets. To overcome this problem, some techniques are proposed. However, the proposed techniques are based on popular SNMP agent and RMON technology, which are characterized by centralization, some inherent known problems are the lack of scalability, complexity to configure, network congestion and not strong local processing ability. This paper proposes a new method for fault detection and isolation based on intelligent mobile agents. To measure the impact on fault detection capability, a monitoring and fault detection experiment is successfully implemented, and the authors compare the number of symptoms detected and traffic of the system with a traditional system using RMON agents. Experimental results show that the fault detection capability using the proposed approach is significantly improved.%随着网络规模的急剧扩大和结构的日趋复杂化,网络故障管理越来越重要.在一个复杂的通信网络中,故障是不可避免的,但是对故障的及时探测和识别对于提高网络的可靠性是非常重要的.监测数据包是网络故障检测中常用的方法,但是在大规模网络中会产生巨量的数据包.为此,人们提出了几种方法.但是这些方法都是建立在集中式管理体系结构之上的,因而没有良好的可扩展性、灵活性和本地处理能力.针对这些问题,提出了一种基于智能移动代理的网络故障检测系统结构,并用Java和aglet实现.实验表明,这种系统结构对于网络故障的检测是非常有效的.
展开▼
