Task Collaborative Offloading for UAV-Assisted Edge Computing With Dynamic Pricing

Abstract

With the demand for real-time data processing in mobile environments has surged, uncrewed aerial vehicle (UAVs) are regrading as flying base stations (BSs) for real-time application and emergency communication. In this article, we investigate a task collaborative offloading in UAV-assisted edge computing environments, integrating dynamic pricing mechanisms and UAVS group formation to optimize resource allocation. We explore the challenges posed by the heterogeneity of UAVs and the dynamic workload distribution. Our proposed system leverages a multiagent deep reinforcement learning framework to intelligently assist computing UAVs to form a collaborative group, considering the constraints of latency, service budget, and computational capacity. The dynamic pricing model incentivizes leading UAV to help efficient task offloading by task collaborative scheduling within groups and task relaying to BS. Through extensive simulations, we demonstrate that our approach significantly enhances the overall system performance, reduces task completion time, and optimizes resource utilization.