Securing IoT-SDN Models: A Comprehensive Review of Deep Learning Approaches and Challenges
P. Navaneethakrishnan, Smitha Elsa Peter · 2024
Rapid Advancements of Internet of things (IoT) devices and the integration of Software-Defined Networking (SDN) have introduced complex security challenges that demand innovative solutions. This survey paper presents a critical analysis of recent advancements in deep learning (DL) models aimed at enhancing the security of IoT-SDN environments. By reviewing a series of research literature, the primary threats faced by the network, including unauthorized data access, service disruptions, and privacy breaches are analysed. This research study highlights several DL techniques, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Deep Belief Networks (DBNs), which have been effectively applied to detect and mitigate such threats. This study assesses various models based on their technique, utilized datasets, evaluation metrics with numerical values, and presents a comparative analysis to describe their merits and limitations. The research findings suggest that while DL models offer significant improvements in detecting complex attack patterns compared to traditional methods, challenges related to model interpretability, resource constraints, and adaptive threat landscapes persist. This research study also discusses about future research directions by highlighting the need for scalable, efficient, and dynamic DL-based security solutions for the evolving IoT-SDN scenarios.