Enhancing intrusion detection performance in industrial internet of things using modified hyperparameter optimization

Abstract

Many sectors have seen significant improvements in operational efficiency and modern connection due to the increasing usage of industrial internet of things (IIoT) devices, a phenomenon known as IIoT. However, efficient intrusion detection systems (IDS) are essential for IIoT security because this quick growth also makes the network more susceptible to sophisticated cyberattacks. The high dimensionality and dynamic nature of IIoT data frequently cause traditional IDS algorithms to struggle, which results in insufficient threat detection and response. This paper uses a novel technique called Bayesian hyperparameter optimization with decision trees to enhance IDS performance in complex IIoT scenarios. Using the large IIoT-23 dataset, the method achieved an excellent 96.07% accuracy rate and a weighted average F1-score of 96.03%, in addition to demonstrating a significant increase in identifying a range of attack scenarios. In direction to growth the precision and robustness of our intrusion detection algorithms, this invention emphasizes how important it is to carefully modify complex hyperparameters and combine decision trees with ensemble methods like genetic algorithms and Bayesian optimization. The results show how successfully the approach adapts to budding threat landscape, expressively lessens false positives, and efficiently handles the enormous volumes of data that are a part of IIoT networks.