An ID2T Toolkit-Generated Synthetic Dataset for AI-Based Intrusion Detection in iV2I Networks

The rise of Industrial Vehicle-to-Infrastructure (iV2I) communication has brought unprecedented efficiency and automation to environments such as warehouses, construction sites, and smart factories. However, these advancements also introduce significant cybersecurity risks, including network-level attacks that can disrupt safety-critical operations. This paper presents an AI-driven Intrusion Detection System (IDS) specifically tailored for industrial iV2I networks. Leveraging the Intrusion Detection Dataset Toolkit (ID2T), we generate a novel synthetic dataset by injecting realistic attack traffic such as DDoS, PortScan, and MemCrash exploits into benign iV2I communication traces. A Multi-Layer Perceptron (MLP) neural network is then trained on this enriched dataset using carefully engineered features extracted from packet flows. The proposed IDS demonstrates high detection accuracy across practical, balanced, and threat-centric scenarios, validating the effectiveness of synthetic datasets in simulating domain-relevant threats. In addition to strong classification performance, the system highlights how synthetic datasets can be used as an alternative to real-world data by significantly reducing the effort and resources required to simulate genuine cyberattacks. This work not only contributes a scalable and reproducible dataset but also emphasizes the importance of AI models in detecting network threats with regard to iV2X deployments.