Intrusion Detection Using Variable-Length Audit Trail Patterns

摘要: Audit trail patterns generated on behalf of a Unix process canb e used to model the behavior. Most approaches proposed so far use table fixed-length represent model. However, variable-length seem be more naturally suited behavior, but they are also difficult construct. In this paper, we present novel technique build patterns. This is based Teiresias, an algorithm initially developed for discovering rigid in unaligned biological sequences. We evaluate quality our testbed environment, and compare it with intrusion-detection system by Forrest et al. [8], which The results achieved method significantly better than those obtained original