The rise of connected vehicles and intelligent transportation lead to the emergence of novel complex risks. Of particular concern is the potential for large-scale attacks to disrupt road transportation, which is the lifeline of cities. This concern has only been growing with the increase in cybersecurity incidents and disinformation attacks in related infrastructures. In this study, we develop a framework to quantify, detect, and mitigate cascading consequences of attacks on road transportation networks. Application of our framework to the road network of Boston reveals that targeted attacks on a small fraction of nodes leads to disproportionately larger disruptions of routes. We develop an unsupervised machine learning algorithm based on network percolation theory and density based clustering (P-DBSCAN) to quantify risk for urban networks based on real-time traffic data. Our study illustrates a holistic approach to build resilience in existing road networks to attacks. Finally, we discuss the applicability of our framework in other smart city infrastructures.

互联汽车和智能交通的兴起导致新的复杂风险的出现。特别令人担忧的是，大规模袭击有可能破坏作为城市生命线的道路交通。随着相关基础设施中网络安全事件和虚假信息攻击的增加，这种担忧只会越来越大。在这项研究中，我们开发了一个框架来量化、检测和减轻道路交通网络攻击的级联后果。我们的框架在波士顿道路网络中的应用表明，对一小部分节点的针对性攻击会导致不成比例的更大的路线中断。我们开发了一种基于网络渗透理论和基于密度的聚类 (P-DBSCAN) 的无监督机器学习算法，以根据实时交通数据量化城市网络的风险。我们的研究展示了一种在现有道路网络中建立抵御攻击能力的整体方法。最后，我们讨论了我们的框架在其他智慧城市基础设施中的适用性。