Dataset of "Passive Identification of PLC Models in Industrial Networks via MAC Address Analysis"

Device identification in industrial networks is crucial for ensuring cybersecurity, supporting penetration testing, and enabling asset management in Operational Technology (OT) environments. This paper presents a novel approach for identifying Siemens PLC device models based solely on MAC addresses and their statistical characteristics. We constructed a custom dataset of 81 MAC address samples from four Siemens PLC models by combining data from laboratory hardware and open sources. From these addresses, we derived six simple yet informative statistical features, including the distance from the mean, median, and z-score of the MAC ID segment. These features were used to train a feedforward neural network, achieving 80\% accuracy on an independent test set, outperforming traditional models such as XGBoost. Our method is passive, rapid, and non-intrusive, making it particularly suitable for sensitive industrial environments. This study also discusses the advantages and limitations of MAC-based identification and proposes scalable strategies for expanding the dataset using publicly available information.