Adaptive Cyberattack Detection in IoT-Edge-Cloud Environments Using Decision Tree Regressor
G C Shwethashree, S H Manjula · Engineering Technology & Applied Science Research · 2025
By facilitating smooth communication among smart devices, edge computing, and cloud environments, the Internet of Things (IoT) has reshaped several sectors. However, IoT networks are highly vulnerable to cyberattacks, particularly link attacks, which compromise security. In the last ten years, various existing Machine Learning (ML) and Deep Learning (DL) approaches have been presented for attack detection, but they often fail to maintain high accuracy when attack patterns evolve. This study proposes a Decision Tree Regressor (DTR) model for attack prediction and adaptation in an IoT Edge-Cloud environment. The model is implemented using SENSORIA for IoT data collection and CloudSim for edge-cloud simulation to efficiently detect attacks. The DTR model dynamically adapts to changes in attack behavior through statistical monitoring. The model was evaluated on the ToN-IoT and UNSW-NB15 datasets, achieving 99.92% and 99.96% accuracy, respectively, significantly outperforming existing approaches. The results demonstrate that DTR improves the accuracy of attack detection while adapting to evolving attack patterns, ensuring robust IoT security.