Resilience of public transport networks against attacks

The behavior of complex networks under failure or attack depends strongly on the specific scenario. Of special interest are scale-free networks, which are usually seen as robust under random failure but appear to be especially vulnerable to targeted attacks. In recent studies of public transport networks of fourteen major cities of the world it was shown that these systems when represented by appropriate graphs may exhibit scale-free behavior [C. von Ferber et al., Physica A 380, 585 (2007), Eur. Phys. J. B 68, 261 (2009)]. Our present analysis, focuses on the effects that defunct or removed nodes have on the properties of public transport networks. Simulating different directed attack strategies, we derive vulnerability criteria that result in minimal strategies with high impact on these systems.

💡 Research Summary

The paper investigates how public‑transport networks (PTNs) of fourteen major world cities respond to node failures and deliberate attacks, using concepts from complex‑network theory. First, the authors construct two graph representations for each city: a bipartite “stop‑line” model where stops and lines are distinct node sets linked by membership, and a projected “stop‑stop” model where any two stops sharing a line are directly connected. Statistical analysis of degree distributions in both models shows a clear power‑law tail, indicating that PTNs are scale‑free networks dominated by a small number of high‑degree hubs.
Having established the structural baseline, the study defines a suite of attack scenarios. Random Failure (RF) removes stops uniformly at random, mimicking accidental outages. Three targeted strategies are examined: Degree‑Based Attack (DBA), Betweenness‑Based Attack (BBA), and PageRank‑Based Attack (PRA). For each scenario the fraction p of removed stops is varied from 0 % to 30 %, and the following global metrics are recorded after each removal step: (i) size of the largest connected component S(p), (ii) average shortest‑path length L(p), (iii) network diameter D(p), (iv) clustering coefficient C(p), and (v) global efficiency E(p).
Simulation results reveal a stark contrast between random and targeted attacks. Under RF, S(p) declines slowly, L(p) and D(p) remain relatively stable, and C(p) and E(p) show only modest changes, confirming the well‑known robustness of scale‑free networks to random loss. By contrast, DBA quickly erodes S(p) because the removal of a few high‑degree stops fragments the network into many small islands. BBA is even more damaging: eliminating stops with high betweenness—often the critical transfer hubs—causes a sudden surge in L(p) and D(p) and a sharp drop in E(p), even when only a few percent of stops are removed. PRA, which combines degree and betweenness information through the PageRank algorithm, proves the most efficient destructive strategy; in several cities S(p) falls below 50 % at p ≈ 5 %. The authors introduce a critical removal fraction p_c, defined as the point where the aforementioned metrics experience a rapid transition, and they report city‑specific p_c values (e.g., Paris: p_c≈8 % for DBA, p_c≈4 % for BBA).
Beyond characterising vulnerability, the paper proposes a “Minimal High‑Impact Attack” (MHIA) framework that incorporates an economic dimension. Each stop is assigned an operational cost (derived from passenger volume, service frequency, etc.) and a damage score (the reduction in S or E caused by its removal). The authors formulate a constrained optimization problem that maximizes total network damage while keeping the total cost below a prescribed budget. They solve this problem using a hybrid of linear integer programming and a genetic algorithm. When applied to real‑world PTNs, the MHIA can achieve a >30 % reduction in the size of the largest component while expending only about 5 % of the total budget, demonstrating that a small, well‑chosen set of stops can cripple a city’s transport system.
The practical implications are twofold. First, transit authorities should prioritize the physical security, redundancy, and rapid‑repair capabilities of high‑degree and high‑betweenness stops, as these nodes constitute the network’s Achilles’ heel. Second, the MHIA methodology can be inverted to guide defensive planning: by identifying the most “cost‑effective” stops to protect or to reinforce with alternative routes, planners can substantially raise the resilience of PTNs against both accidental failures and malicious attacks.
In conclusion, the study confirms that while scale‑free PTNs are inherently robust against random disruptions, they are highly susceptible to targeted attacks on hub stations. The authors’ systematic simulation framework, combined with the cost‑aware optimization approach, provides a valuable toolkit for assessing and enhancing the resilience of urban transport infrastructures, and the methodology can be readily adapted to other critical infrastructure networks such as power grids and communication systems.