Nowadays with the fast development of IT’s technologies, new services and applications improved people’s daily life. They are supported by tons of devices that are continuously sharing huge and heterogeneous data. In this challenging scenario, security issues are increasing and several threats arise where network communications and systems are targeted for attacks. In order to counteract against them, new tools and methods need to be proposed. The Multivariate Statistical Network Monitoring (MSNM) is a promising methodology for anomaly detection as demonstrated in several works. In this work, the practical application of this methodology is tested by means of the tool called MSNM-Sensor. Its detection performance is evaluated in common and harmful network attacks included in recently built network datasets. In particular, authentication based and different Denial of Service attack (DoS) types are successfully detected by MSNM-Sensor as two of nowadays relevant and harmful security threats. In comparison to similar state of the art approaches, the MSNM-S outperforms them in almost all the types of DoS and Brute Force attacks considered in this work.

如今，随着IT技术的快速发展，新的服务和应用程序改善了人们的日常生活。大量不断地共享巨大且异构数据的设备为它们提供了支持。在这种具有挑战性的情况下，安全问题日益严重，在网络通信和系统成为攻击目标的情况下，出现了多种威胁。为了对抗它们，需要提出新的工具和方法。多项统计网络监测（MSNM）是一种有前途的异常检测方法，如多项研究所示。在这项工作中，此方法的实际应用是通过称为MSNM-Sensor的工具进行测试的。在最近建立的网络数据集中包含的常见和有害网络攻击中评估了其检测性能。特别是，MSNM-Sensor已成功将基于身份验证和不同的拒绝服务攻击（DoS）类型检测为当今两种相关且有害的安全威胁。与类似的现有技术相比，MSNM-S在本文研究的几乎所有类型的DoS和蛮力攻击中均胜过它们。