Connected Vehicle-enabled Centralized Routing with Dynamic Network Allocation

The Macroscopic Fundamental Diagram (MFD) enables effective and cost-efficient network traffic management to gain system optimum. The recent emergence of connected vehicles (CVs) opens avenues to leverage MFD for centralized management, further enhancing urban network mobility. This paper aims to augment an existing MFD-based centralized routing system with dynamic locality allocation. Employing the Girvan-Newman algorithm, the system divides the network's localities into smaller regions, each reflecting distinct traffic conditions, computing resource availability, and management efficiency. This approach enhances traffic management and optimizes resource utilization, ultimately improving system performance. Centralized routing, based on dynamic localities, seeks optimal CV paths, reducing computational costs and achieving network optimization. Evaluation results validate the centralized system's advantages in enhancing urban mobility across various CV Market Penetration Rates (MPRs) and evolving traffic conditions. The dynamic locality allocation system surpasses static locality counterparts, particularly ex-celling at high MPRs. Notably, in scenarios with very high MPRs, the centralized system slightly sacrifices CV mobility benefits to enhance overall network performance. Overall, this enhancement is essential in elevating network-wide vehicle performance, ensuring smoother traffic flow, and mitigating congestion levels.