Engineering edge-oriented digital twin networks for smart cities

Driven by rapid urbanization, Smart Cities are emerging as interconnected ecosystems that leverage advanced technologies to enhance urban living. The growing diversity of urban services and citizen needs highlights the urgent demand for transformative solutions capable of managing the increasing complexity of urban environments. To address these challenges, Digital Twins (DTs) have emerged as a key enabling technology for Smart Cities. By creating virtual replicas of city assets and processes, DTs support closed-loop adaptive systems for analyzing city dynamics and enabling (semi-)autonomous optimization strategies. While early DT deployments were primarily cloud-based, recent advances have shifted toward a resource continuum that incorporates edge computing, giving rise to Digital Twin Networks (DTNs). Hosting fine-grained models closer to end users improves responsiveness and helps meet stringent operational requirements of critical city services. This dissertation investigates the role of DTNs in Smart City contexts, analyzing their applicability and deriving key design requirements. Its primary contribution is the definition of a research roadmap that structures and guides the development of DTN-supporting solutions, with emphasis on edge infrastructures. Along this roadmap, the thesis proposes and evaluates methods and tools addressing the main challenges associated with DTN realization. Specifically, this work contributes: (i) a scalable edge-based approach for near-real-time traffic simulation that combines automated scenario generation from vehicular data with distributed execution and adaptive load balancing to reduce simulation time; (ii) approaches enabling application-aware networking to support heterogeneous and dynamic Quality of Service requirements of DTN services; and (iii) a solution for large-scale DTN service deployment at the edge, extending the Multi-Access Edge Computing architecture to leverage far-edge resources, particularly vehicles. All solutions have been thoroughly validated, highlighting both advantages and limitations. Overall, this framework lays the foundation for efficient and scalable DTNs capable of supporting the continuous evolution and growing complexity of modern urban environments.