Detection of Distributed Denial-of-Service (DDoS) Attack with Hyperparameter Tuning Based on Machine Learning Approach

Distributed Denial-of-Service (DDoS) attack is a malicious cyber-attack which targets availability element in CIA triad and to disrupt the availability of network services of a target by performing a huge malicious traffic flood. To conduct the study, a standard benchmark dataset DDoS Attack SDN Dataset is applied. EDA and Data Pre-processing are performed to ensure a clean dataset is produced for obtaining an accurate and meaningful detection performance results. Hyperparameter tuning is performed to enhance the detection performance of the models. It is proposed that DNN shows the promising results as it has shown 99.84% accuracy to detect DDoS attack after performing hyperparameter tuning. It is observed that hyperparameter tuning has improved and increased most of the performance results of DNN and DT, with increment 4.84% in DT while 0.97% in DNN. Besides, the detection results have been increased and their false detection has been reduced. This study could help to reduce the dwell time of DDoS attack, increase the Mean Time To Contain (MTTC) and avoid alarm fatigue.