Resilience Evaluation of Smart City Network Topology Graph Generator: A Real‐World Network Analysis Approach

ABSTRACT Smart cities rely on resilient network topologies for reliable service delivery, yet tools for analyzing them remain limited. This study evaluates our Smart City Network Topology Graph Generator (SCGG), which creates pseudo‐random topologies with the multilayered, clustered structure of smart city networks. SCGG is characterized through a conceptual comparison with the generators NetDiff and DGGI and evaluated against two real deployments, the Manado e‐Government Network and the Toulouse Public Transportation Network, using centrality, reliability, and performance metrics. The SCGG topologies preserved the minimal node and edge connectivity profile of both real networks, demonstrating a correspondence at the minimal connectivity level, with robustness assessed by the attack simulations. The generated degree distributions differ significantly from the real ones (Kolmogorov–Smirnov statistics of 0.53 and 0.49 against a critical value of about 0.28), so distributional fidelity is traded for redundancy. At natural operating densities, SCGG improved network efficiency by 10% and 21% and shortened average path lengths by 9% and 20% for Toulouse and Manado, respectively. Under targeted attacks, the generated topologies retained a markedly larger connected component, and a density‐controlled comparison attributes this tolerance and the clustering gain to the deliberately added redundant edges. Even at an equal edge budget, SCGG achieved 39% higher efficiency and 40% shorter paths for Toulouse. This advantage reflects the logical structure of the generator, which is not subject to the geographic embedding of the deployed networks. These findings, grounded in repeated independent runs, position redundant connections and evenly distributed communications as key requirements for resilient smart city infrastructure.