Adaptive Load-Aware Offloading for UAV-Assisted IoT Edge Computing in Low-Altitude Economy Scenarios

Abstract

The low-altitude economy is accelerating the deployment of unmanned aerial vehicles (UAVs) as flexible sensing, communication, and edge-computing platforms. In UAV-assisted Internet of Things (IoT) mobile edge computing (MEC), conventional greedy offloading can become unreliable because per-task decisions ignore shared wireless bandwidth, edgecomputing contention, and deadline heterogeneity. This paper proposes an Adaptive Load-Aware Offloading (ALAO) algorithm for a three-tier IoT-UAV-MEC system. ALAO constructs an urgency-aware and load-refined candidate schedule considering task delay, energy consumption, node load, and deadline-violation risk, and then compares this candidate with a conventional greedy schedule using a completion-oriented lexicographic utility. Simulations averaged over 20 random seeds show that ALAO remains comparable to Greedy under the default setting, where the completion-rate improvement is only 0.19 percentage points, but provides clearer gains under resource-constrained and highdensity conditions. In particular, when the MEC CPU frequency is reduced to 5 GHz, ALAO improves the task completion rate from 21.75% to 31.36% and reduces average delay by 16.96%. Under N = 80 IoT devices, ALAO improves completion rate from 46.04% to 48.22% and reduces average delay by 4.00%. These results indicate that ALAO should be viewed as a lightweight contention-oriented offloading method for resource-constrained low-altitude IoT-MEC systems, rather than as a universally superior policy under all parameter settings.