Dynamical interaction network in urban traffic

Introduction: Urban traffic systems transition dynamically between congestion and free-flow states, driven by local interactions between road segments or regions. Understanding how these interactions contribute to congestion, including system-wide congestion, is crucial for effective traffic management. However, existing research has overlooked the dynamical nature of these interactions, which are essential for capturing the changing behavior of urban traffic. Methods: In this study, we use a pairwise maximum entropy model to infer interaction networks from sliding time windows and analyze their dynamics during typical daily periods: morning peak, noon off-peak, and evening peak. Results: We find that (1) interaction networks remain stable within each period but exhibit structural shifts between periods, especially between peak and off-peak periods; (2) stable high-strength edges in dynamical interaction network are characterized by long-range and negative interactions; (3) the proportion and modularity of positive interactions, along with the strength of negative interactions, are important structural features that distinguish peak from off-peak hours. Discussion: These results provide new insights into how local interaction dynamics drive global state transitions in urban traffic, offering guidance for improving traffic resilience through targeted control strategies.