Rajeh Wahid, Aborokbah Majed, S Manimurugan, Albalawi Umar, Aljuhani Ahamed, Younes Osama Shibl Abdalghany, Periyasami Karthikeyan
Faculty of Computers and Information Technology, University of Tabuk, Tabuk, Saudi Arabia.
School of Computer Science and Engineering, RV University, Bengaluru, Karnataka, India.
PeerJ Comput Sci. 2025 Mar 31;11:e2743. doi: 10.7717/peerj-cs.2743. eCollection 2025.
Smart cities, enabled by the Internet of Things (IoT), leverage technology to optimize urban living and enhance infrastructure. As urban environments become interconnected hubs of digital innovation, securing critical components like public transportation infrastructure becomes increasingly important.
This research addresses the need for robust intrusion detection systems (IDS) tailored to the unique challenges of securing public transportation within smart cities. Focused on the Tabuk region in Saudi Arabia, the study introduces an IDS model integrating the deep maxout network with walrus optimization (DMN-WO). The DMN is configured with an architecture that includes multiple layers with maxout activation functions. These layers are capable of capturing complex patterns in the data, making the DMN particularly effective for identifying anomalies in IoT network traffic. The DMN-WO model is ensured to be resource-efficient and suitable for real-time deployment on constrained devices like Raspberry Pi, typical in IoT systems.
Training and validation are conducted using the CIC-IDS-2018 dataset, CIC-IDS -2029 dataset and real-time data from Raspberry Pi devices deployed in the smart city's public transportation network. Real-time data application maintains robust performance, with 98.06% accuracy, 98.50% detection rate, 98.81% precision, 98.24% specificity, and a 98.57% F1-score.
This research advances cybersecurity measures in smart city applications by providing a resilient solution for detecting and mitigating security threats in public transportation infrastructure. It lays the groundwork for further refinements and real-world deployments in the dynamic landscape of smart cities.
由物联网（IoT）赋能的智慧城市利用技术优化城市生活并加强基础设施建设。随着城市环境成为数字创新的互联枢纽，保障公共交通基础设施等关键组件的安全变得愈发重要。
本研究满足了针对智慧城市中保障公共交通安全的独特挑战量身定制强大入侵检测系统（IDS）的需求。该研究聚焦于沙特阿拉伯的塔布克地区，引入了一种将深度最大输出网络与海象优化算法（DMN-WO）相结合的IDS模型。DMN采用包含多层且带有最大输出激活函数的架构进行配置。这些层能够捕捉数据中的复杂模式，使得DMN在识别物联网网络流量异常方面特别有效。确保DMN-WO模型具有资源高效性，适合在物联网系统中常见的如树莓派等受限设备上进行实时部署。
使用CIC-IDS-2018数据集、CIC-IDS -2029数据集以及部署在智慧城市公共交通网络中的树莓派设备的实时数据进行训练和验证。实时数据应用保持了强大的性能，准确率为98.06%，检测率为98.50%，精确率为98.81%，特异性为98.24%，F1分数为98.57%。
本研究通过提供一种用于检测和缓解公共交通基础设施安全威胁的弹性解决方案，推进了智慧城市应用中的网络安全措施。它为在智慧城市的动态环境中进行进一步优化和实际部署奠定了基础。
