ML-Assisted Security for the Detection of DDoS Attacks in Connected IIoT Environment: Implementation and Comparative Analysis

Distributed denial-of-service (DDoS) attacks are cyber security malicious attempts to disrupt the regular traffic of an Industrial Intenet of Things (IIoT) network service by flooding the traffic of the connected IIoT infrastructure with endless packets. The article discusses DDoS attacks in connected IIoT environment and the use of machine learning (ML) models for defend against such security malicious attempts. For that, we initially explore the different types of DDoS attacks, the vulnerabilities of the network layer, and various ML techniques used for detecting attacks. Different ML-assisted security models are implemented and analyzed for detecting DDoS attacks targeting IIoT network. These models include support vector machine, logistic regression, naive Bayes, and neural network to learn the attack detection. Comparisons are made with these ML models on behalf of the accuracy and execution time. Further, we implement parallel processing to reduce the execution time of the training part, and comparisons are made considering different environments, CPU, and GPU. Our analysis of the results demonstrates a significant reduction in training time for the models while maintaining accuracy when using parallel processing to implement the ML-assisted security models.