Time varying topology adaptation for internet of vehicles task offloading

The rapid growth of in-vehicle applications has driven the adoption of edge computing in the Internet of Vehicles, enabling resource-constrained vehicles to offload tasks for efficient execution. However, this optimization is challenging due to dynamic network conditions and the need to balance latency, energy, and task completion, making real-time optimal solutions intractable. To address these issues, this paper investigates a heterogeneous computing architecture consisting of task vehicles, idle serving vehicles, and roadside units, and formulates a multi-objective optimization model that jointly minimizes system latency, energy consumption, and task dropping rate. To tackle the decision-making challenges arising from vehicle mobility, channel fluctuations, and dynamic node load variations, a Time-varying Topology Adaptation Algorithm is proposed. The algorithm incorporates a dual adaptive mechanism: an internal parameter adaptation module that dynamically balances the exploration and exploitation capabilities of the algorithm, and an external environment-aware adaptation mechanism that integrates real-time multi-dimensional states to align heuristic rules with the dynamic network environment. Theoretical analysis verifies the convergence of the algorithm under dynamic constraints. Finally, comparative experiments on real-world datasets validate the effectiveness of the proposed algorithm in improving system utility, latency, energy consumption, and task dropping rate under high-mobility and high-load conditions. The source code is publicly available at https://github.com/Lose-Code/TTA .