Advancing Traffic Resource Scheduling With Cloud–Edge Collaboration: A Virtualized Digital Twin Perspective

With the continuous advancement of urban transportation digitalization, the resources confronted by intelligent transportation system are exhibiting increasing diversity and scale, leading to a growing contradiction between resource demand and scheduling efficiency. To further enhance the operational efficiency of transportation systems, this article proposes a virtualization-based digital twin transportation resource scheduling model under a cloud–edge collaboration framework. This model leverages the characteristics of digital twin technology to achieve efficient resource operation through virtualized resources and a cloud–edge collaborative architecture, adapting to on-demand scheduling in various complex traffic scenarios. Specifically, this article integrates digital twin technology to perceive and map the physical and digital spaces of transportation through a virtual-real fusion approach. It employs a multiresource collaborative scheduling architecture based on cloud–edge collaboration to achieve effective allocation of transportation resources. Furthermore, it utilizes a virtualized resource representation to establish efficient and unified integrated resource management based on a digital global view. Additionally, to further optimize scheduling, this article proposes a value distribution policy gradient graph deep reinforcement learning model based on node importance strategy. Through deep learning of key nodes and relationships and adaptive feedback to scheduling demands, it achieves efficient collaboration and scheduling of digital cloud–edge integrated transportation resources. Experimental results demonstrate that the proposed model outperforms current mainstream resource scheduling models in terms of resource utilization, long-term average revenue, and revenue-cost ratio.