Deep Learning-Based Echo State Neural Network for Cyber Threat Detection in IoT-Driven IICS Networks

Abstract

The advent of Software-Defined Networking (SDN) has ushered in a new era in network architecture, providing unprecedented levels of flexibility and adaptability. However, this advanced flexibility exposes SDN to security risks, particularly Distributed Denial of Service (DDoS) attacks. Detecting and mitigating DDoS attacks in SDN environments poses a critical challenge. This study introduces an innovative DDoS detection approach leveraging Echo State Networks (ESN) tailored specifically for SDN. This approach is based on two core assumptions: firstly, routine network operations primarily exhibit normal behavior, and secondly, there are discernible differences in data characteristics between normal and abnormal network conditions. These assumptions hold true in the realm of everyday network dynamics. To validate the efficacy of the ESN algorithm, we augment this approach by incorporating flow features to enhance DDoS detection capabilities. This study underscores the effectiveness of ESN in identifying and mitigating Distributed Denial of Service (DDoS) attacks, DDoS threats, achieving an impressive average success rate of 97.78%. By harnessing the potential of Echo State Networks, this work makes a substantial contribution to ongoing efforts in fortifying network security, providing a proactive defense against disruptive DDoS attacks.