On the Performance of Reactive and AnticipatoryFault Management Agents in Adaptive Networks

摘要

In this paper we develop an intelligent agent-based networkrnmanagement model, which uses an anticipatory agent tornproactively detect occurrence of faults using a predictivernmodel pertaining to network performance. Two reactivernrule based fault management strategies are comparedrnagainst the anticipatory approach. Our preliminary findingsrnindicate that the anticipatory technique improves networkrnperformance significantly better than the reactiverntechniques. We furthermore describe a methodology forrnadaptive restructuring of the network based on thernsimulated annealing process. We observe that adaptivernrestructuring gives significantly better performance underrnthe reactive rule-based fault-management technique asrncompared to the anticipatory strategy.