Frame of Interest Approach on Quality of Prediction for Agent-Based Network Monitoring

Abstract

We present an approach to compute the quality of prediction for network monitoring. The monitoring is part of a proactive mobile agents based management system for network health (magmaNH). To allow prediction of a system's behavior, magmaNH contains prediction services placed on core nodes of a network. To make predictions as precise as possible, a measure and a process have to be defined, which enable to determine the quality of predictions. This measure of quality enables magmaNH optimizing the prediction services to become a reliable support system for automated network management. © Springer-Verlag 2004.