Towards a Framework for Improving Experiments on DoS Attacks

In recent years, a number of solutions have been proposed for the detection of Denial of Service (DoS) attacks. Most of them have been tuned and tested by means of publicly available labelled datasets, which can be conveniently used to overcome the scarceness of real-life data gathered under incidents and attacks from production environments. Notwithstanding the high detection rates of existing algorithms, there is little concern about the representativeness of public traffic data and the impact on continuity of operation of the victim services.