Abstract

Due to the emergence of online examination systems, there have been significant cybersecurity issues, particularly in Distributed Denial of Service (DDoS) attacks, that may cause disruption of system availability and compromise the integrity of the assessments. This work promotes a multi-layered approach to mitigation to ensure that secure, reliable, and uninterrupted examination services are provided during adversity. The architecture proposed combines traffic filtering, anomaly detection (with a machine learning element), adaptive rate limiting, dynamic load balancing, and cloud-based traffic scrubbing into one line of defence. In training and testing the detection model, a dataset containing both benchmark traffic and simulated examination-specific network traffic was used in order to provide a comprehensive examination of normal and malicious network traffic patterns. Experimental results show that the model can achieve a detection rate of 97% at a low false positive rate and high traffic, 520 Mbps. Under normal conditions, the system can achieve an average response time of 150ms, which is quite suitable for controlling latency, even under a heavy attack scenario. Furthermore, the framework reduces the drop rate of packets down to 4% and increases the overall service availability, reducing the downtime by an order of 41% as compared to the traditional mitigation measures. Ablation study also shows that the intelligent detection and adaptive mitigation functions can significantly enhance the performance and robustness of the system. The results indicate that there is a framework that can successfully address various attack vectors in DDoS attacks with a relatively low computation cost and without disrupting the functioning of the system. The study reveals the power of using machine learning to identify attacks and scalable network defence strategies to improve resilience. The proposed solution promises the prospect of enhancing the security and integrity of the online examination platform and, in the same breath, maintains its continuity and validity to the online assessment ecosystem, which is increasingly becoming a reality. The solution is not only viable but scalable as well and might be implemented to improve the security, integrity, and continuity of online exam platforms so as to provide a fair and reliable online assessment environment.