The constant growth of Denial of Service (DoS) attacks stands as a significant threat in our digital and ultra-connected world. Their danger is enhanced by the difficulties in their detection. In fact, most of the various detection methods do not provide an exhausting and immediate solution to the problem. In particular, the detection of slow attack results problematic and requires a considerable human effort. Modern solutions have opted for an artificial intelligence (AI) oriented approach, which consist of creating of a model trained with a representative dataset, capable of recognising hardly detectable patterns and doing so automatically. Furthermore, the necessity to keep note of traces and proof of the attacks emerged, because they may be lost in the tentative of rebooting the system.