ALNS-DR: An Optimization Method for Large-Scale Task Offloading in UAV-Assisted Edge Computing

The integration of Unmanned Aerial Vehicle (UAV) and edge computing technology extends the coverage of multi-access edge computing networks, but it also introduces complex-ities in task offloading and resource allocation. While current research predominantly focuses on optimizing small-scale task offloading, its impact on large-scale task offloading remains limited. To bridge this gap, we present a large-scale task offloading model. Specifically, we consider that each terminal equipment is capable of offloading tasks to other terminal equipment and UAV-assisted edge servers that can be connected to it. Additionally, we present an adaptive large neighborhood search algorithm based on dynamic random operator to achieve an optimal solution. The operator combines the Metropolis algorithm with the roulette wheel selection algorithm, allowing for adaptive weight adjustments during the selection of destroy/repair operators, thereby enhancing the algorithm's ability to escape local optima during the convergence process. The simulation results confirm the efficacy of the proposed algorithm in significantly reducing system energy consumption during large-scale task offloading scenarios.