Development of an adaptive learning network-failure detection system

The purpose of this work consists in finding a method able to detect anomalies and adapt to new behaviours in an IP-network. Here is possible to find a work of investigation about the different anomaly detection systems. During the introduction and the first points, the audience could realize how the field of anomaly detection is divided in two big groups: Anomaly detection systems based in rules, and adaptive anomaly detection systems. These two ideas will be discuss, and some examples of each technology are given. In the case of adaptive anomaly detection, there are some techniques proposed. One of them is developed during the rest of the work. I wanted to implement this technique in a small network property of the RUS department of the University of Stuttgart. The implementation, the problems found and additional information can be found in this report. Later the tests and results applied make us think if the behaviour is correct or not. The discussion and the ideas obtained can be found at the end of the report, where is possible to find also the advices for future developers. This report has been designed as a guide for developers which lend them to avoid too much time in understanding the present algorithm. The aid of this work is to be considered as a helper reference to people who wants to implement this kind of systems.Ingeniería de Telecomunicació