Guardians of the IoT: A Symphony of Ensemble Learning for DDoS Attack Resilience
Aadil Khan, Ishu Sharma · 2023
DDoS attacks flood IoT networks with traffic, interrupting performance and perhaps causing faults, placing smart devices at danger. DDoS attacks on IoT devices may take them offline and interrupt their functioning. Malicious traffic may cause service failures, data breaches, and device damage. These attacks may exploit weaknesses, allowing unwanted access and security breaches in IoT ecosystems. Protecting IoT devices's integrity and performance demands strong cybersecurity techniques. IoT device attacks that use DDoS must be detected early to reduce risks, service interruptions, and damage. Swift identification enables quick countermeasures, protecting networked devices and IoT environments. DDoS attacks must be detected and stopped early to minimize network performance and security risks. In this research paper, machine learning methods are evaluated to identify DDoS attacks on IoT devices using the Aposemat IoT-23 dataset. Sensitivity, F1-score, accuracy, and ROC-AUC are used to investigate the ensemble learning technique. Using the Aposemat IoT-23 dataset, the ensemble machine learning system detected DDoS attacks on IoT devices with robust performance. The method captured actual positive attacks with great sensitivity, resulting in a balanced F1-score and accuracy. The algorithm's adaptability under varied maximum depth settings is also examined, revealing its real-world resilience. The results improve IoT security by improving DDoS detection techniques. The algorithm's discriminatory capacity was shown by ROC-AUC values under different maximum depth settings. These findings demonstrate the algorithm's ability to identify and mitigate IoT DDoS attacks early.