A Lightweight MobileNetV2-Based Intrusion Detection System for Mitigating DDoS Attacks on Edge Devices

The increased use of Internet of Things (IoT) has enabled revolution in industrial workflow at the same time it also opened a backdoor for Distributed Denial of Service (DDoS) threats. This not only down the networks, it creates unavailability. As these units typically have no control over in terms of memory and power to protect themselves, the created malicious traffic spikes cause service interruptions and degraded system availability. A quick alert system, which could distinguish suspicious traffic is appreciable. To counter the growing threat of DDoS attacks in Industrial IoT systems, this study develops an efficient deep learning-driven intrusion detection model utilizing MobileNetV2, which is having the ability to extract dominant features with minimal computational resources. while using very little processing power. The presented study relies on the Edge-IIoT network traffic vectors were used to train and test the model. The integration of DSSTE helped the model focus on challenging samples, thereby reducing overfitting and improving detection capability. The tested results of this experiment show that the proposed model is efficient in various evaluation metrics. The model based on MobileNetV2 was able to identify DDoS attacks correctly 98.84% of the time. In addition to detecting an attack, the proposed system also implements traffic rate limiting, packet filtering, and IP blacklisting to reduce the impact of malicious activity once the attack is detected. This study presents a straightforward way to find and stop the DDoS attacks in Edge-IIoT enabled applications.