Cyberthreats are a permanent concern in our modern technological world. In the recent years, sophisticated traffic analysis techniques and anomaly detection (AD) algorithms have been employed to face the more and more subversive adversarial attacks. A malicious intrusion, defined as an invasive action intending to illegally exploit private resources, manifests through unusual data traffic and/or abnormal connectivity pattern. Despite the plethora of statistical or signature-based detectors currently provided in the literature, the topological connectivity component of a malicious flow is less exploited. Furthermore, a great proportion of the existing statistical intrusion detectors are based on supervised learning, that relies on labeled data. By viewing network flows as weighted directed interactions between a pair of nodes, in this paper we present a simple method that facilitate the use of connectivity graph features in unsupervised anomaly detection algorithms. We test our methodology on real network traffic datasets and observe several improvements over standard AD.
翻译:网络威胁是我们现代技术世界长期关注的一个问题。近年来,利用复杂的交通分析技术和异常检测算法来对付越来越多的颠覆性对抗性攻击。恶意入侵被定义为企图非法开采私人资源的入侵行动,表现为非正常的数据传输和/或异常连接模式。尽管文献中目前提供了大量统计或基于签名的探测器,但恶意流动的地形连接部分被较少利用。此外,现有的统计入侵探测器有很大一部分是基于有监督的学习,以标签数据为依据。通过将网络流动视为一对节点之间的加权直接互动,我们在本文件中提出了一个简单的方法,便利在未经监督的异常检测算法中使用连接图特征。我们用真实网络流量数据集测试我们的方法,并观察到一些与标准自动识别法相比的改进。