Makine öğrenimi kullanarak DDOS saldırısı algılaması için trafik sınıflandırması

Anomaly detection remains a major concern for many Internet players. It is a complex problem, having attracted the interest of both industrial players, such as Internet Service Providers (ISPs), and members of the scientific community. These anomalies, defined as events deviating from the usual behaviour of traffic, can have severe consequences on the services provided. While some anomalies may be the result of configuration errors, it is scans and attacks that are of most interest to digital players We present our problem here, divided into three parts. First, we cover all the things to consider when building a detector. We then address several challenges related to current traffic. Finally, we discuss the specificity of distributed denial of service attacks Thus, this thesis seeks to propose new solutions, taking into account all the components of this problem. we first present the motivations of project, for which our work was carried out. The problem to which we answer is then exposed. then, we introduce our different contributions to answer this problemWe present our problem here, divided into three parts. First, we cover all the things to consider when building a detector. We then address several challenges related to current traffic. Finally, we discuss the specificity of distributed denial of service attacksThus, this thesis seeks to propose new solutions, taking into account all the components of this problem. we first present the motivations of project, for which our work was carried out. The problem to which we answer is then exposed. then, we introduce our different contributions to answer this problem.