Cross-regional cooperative multiple-timescale joint traffic control, signal control, and routing for maximum network throughput

Network-level signal control and routing were independently optimized using centralized algorithms that rely on traffic data from the entire network. These methods cannot guarantee provable network performance under joint control and real-time controller updates. Additionally, they were not integrated with traffic control, which may result in low network performance due to oversaturation. This study proposes single-timescale and dual-timescale joint traffic control, signal control, and routing based on cross-regional cooperation for urban networks of varying scales and communication delays. Urban networks are divided into regions based on road structure and regional functions to narrow down the traffic data required for network-level joint control. Traffic control is implemented collaboratively at entry nodes and inter-regional connection nodes with different timescales to balance the traffic at inter-regional connection nodes via demand allocation and to avoid regional oversaturation by limiting inter-regional traffic transmission rates. Real-time signal control and routing are implemented at intersections to optimize traffic priority and route vehicles efficiently. Both joint control algorithms are fully decomposable and distributed. Using Lyapunov drift functions, this study proves that both the single-timescale and dual-timescale joint control algorithms maximize network throughput, although the latter results in higher average delays. Comparative simulations on a grid-like network and a network with non-uniform link density show that the joint control algorithms improve network throughput by up to 19.3% compared to existing algorithms.