Non-deterministic Event Correlation Based on C-F Model

This paper proposes a non-deterministic event correlation technique for diagnosis problem of end-to-end service in network, which uses an event – fault model based on path domain of events. The technique utilizes a refined heuristic approach to create and update fault hypotheses that can explain these events received by system, and computes these hypotheses’ belief by C-F model method. The result of event correlation is the hypothesis with maximum belief. Simulation shows this approach can get a high accuracy even in the case of low observability events ratio.