Joint Latency-Energy Aware Digital Twin Placement in Heterogeneous Cloud-Edge Networks

In the era of the Internet of Vehicles (IoV), Digital Twins (DTs) serve as a critical bridge between physical vehicles and the digital world, enabling real-time virtualization, simulation, and data-driven decision-making. While Cloud Computing (CC) and Mobile Edge Computing (MEC) provide foundational support for low-latency services, the dynamic nature of IoV environments-characterized by fluctuating resource availability, heterogeneous infrastructures, and stringent quality-of-service requirements-poses significant challenges for efficient DT deployment. To address this, we propose a novel DT system framework tailored for heterogeneous MEC/CC environments, where DTs are dynamically maintained across distributed servers using multisource data collected from vehicular networks. Central to our approach is a DT placement optimization problem that jointly minimizes latency and energy consumption under resource constraints. We design a Distributed Deep Learning (DDL)-based offloading scheme to adaptively optimize DT placement, ensuring scalability and responsiveness to real-time environmental changes. Extensive simulations demonstrate that our solution shows superior performance compared to heuristic benchmarks.