IoT-Defender: A Convolutional Approach to Detect DDoS Attacks in Internet of Things

Abstract

The rapid expansion of the Internet of Things (IoT) has been met with a concurrent rise in cybersecurity threats, particularly distributed denial of service (DDoS) attacks. These attacks pose severe risks to the interconnected networks that form the backbone of various critical infrastructures. Traditional defense mechanisms, primarily tailored for more conventional network settings, fall short in effectively tackling the unique complexities and dynamic nature of IoT environments. Their limitations are evident in the face of the diverse range of IoT devices, the enormity of data generated, and the continually evolving strategies of cyber attackers. To address these challenges, this paper introduces "IoT-Defender," an innovative solution that utilizes Convolutional Neural Networks (CNNs) for the detection of DDoS attacks in IoT networks. This system marks a significant advancement in IoT cybersecurity, leveraging deep learning to analyze and interpret the intricate patterns of network traffic. IoT-Defender is built on a robust CNN architecture, comprising multiple convolutional and pooling layers that work synergistically to extract and process complex features from network traffic data. This architecture enables the system to discern between normal operations and malicious activities with high accuracy. Utilizing the comprehensive CICDDoS2019 dataset for training and validation, IoT-Defender demonstrates remarkable efficacy, achieving a detection accuracy of 99.68% and outperforming traditional security models. This high accuracy underscores the system's capability to adapt to the varied and evolving landscape of IoT networks, making it a scalable and adaptable defense mechanism. IoT-Defender thus emerges as a critical tool in enhancing the resilience and security of IoT infrastructures against the threats of DDoS attacks.