Efficient Machine Learning Algorithms for DDoS Attack Detection

Distributed Denial of Service (DDoS) attack is a widely spread attack that posing a major threat to organizations dependent on online services. DDoS attacks aim to disrupt services by overwhelming servers with fake traffic from multiple sources. Early and effective detection of DDoS attacks is crucial for mitigating their impact. Recently, the most widely used algorithms for detecting DDoS are based-on Machine Learning (ML) and Deep learning (DL). The work in this paper focuses on providing a comparative study between recently ML algorithms that were tested using the CICDoS2019 dataset. The objective of this comparison is determining the most effective ML algorithm for DDoS detection. Based on the comparative study results, it is found that the Gradient Boosting (GB) and the XGBoost algorithms are extraordinarily accurate and correctly predicted the type of network traffic with 99.99% and 99.98% accuracy respectively, in addition to, a low false alarm rate of approximately 0.004 for GB.