Energy-Aware Holistic Optimization in UAV-Assisted Fog Computing: Attitude, Trajectory, and Task Assignment
Abstract
Unmanned Aerial Vehicles (UAVs) have significantly enhanced fog computing by acting as both flexible computation platforms and communication mobile relays. In this paper, we consider four important and interdependent modules: attitude control, trajectory planning, resource allocation, and task assignment, and propose a holistic framework that jointly optimizes the total latency and energy consumption for UAV-assisted fog computing in a three-dimensional spatial domain with varying terrain elevations and dynamic task generations. We first establish a fuzzy-enhanced adaptive reinforcement proportional-integral-derivative control model to control the attitude. Then, we propose an enhanced Ant Colony System (ACS) based algorithm, that includes a safety value and a decoupling mechanism to overcome the convergence issue in classical ACS, to compute the optimal UAV trajectory. Finally, we design an algorithm based on the Particle Swarm Optimization technique, to determine where each offloaded task should be executed. Under our proposed framework, the outcome of one module would affect the decision-making in another, providing a holistic perspective of the system and thus leading to improved solutions. We demonstrate by extensive simulation results that our proposed framework can significantly improve the overall performance, measured by latency and energy consumption, compared to existing mainstream approaches.
- A unified cross-layer framework jointly optimizing attitude, 3D trajectory, task assignment, and resource allocation.
- RL-enhanced FEAR-PID for quadrotor attitude control with improved stability across flight phases.
- ACS-DS trajectory planning with decoupling and safety mechanisms for faster, more reliable convergence.
- Deployment-ready recipes for fast experiments on Microsoft AirSim and NVIDIA Isaac Sim.
- Extensive simulations showing significant reductions in latency and energy compared to strong baselines.
System Overview
We study a UAV-assisted fog computing system where a UAV acts as both a mobile computing node and a communication relay. The framework jointly optimizes attitude control, 3D trajectory planning, task assignment, and resource allocation to reduce overall latency and energy consumption under safety and terrain constraints.
The system consists of a quadrotor UAV, a remote data center in the cloud, and multiple mobile IoT devices distributed in a three-dimensional space. Each computational task can be executed locally on the IoT device, processed in the fog layer by the UAV, or offloaded to the central cloud, with the execution location dynamically determined based on real-time network conditions, energy constraints, and task characteristics. The system operates in discrete timeslots, with tasks arriving dynamically from IoT devices following Poisson processes, requiring adaptive decision-making that accounts for the UAV's battery capacity, maximum speed and altitude constraints, and the varying communication channel conditions.
- Cross-layer coupling: attitude control, trajectory planning, and fog computing decisions affect each other.
- 3D, terrain-aware planning: motion is constrained by safety and varying terrain elevations.
- End-to-end objective: jointly minimize total latency and energy consumption.
Figure 1. UAV-assisted fog computing system model.
Method at a Glance
Our holistic pipeline connects four interdependent modules. The output of each module feeds back into others, enabling end-to-end optimization from UAV attitude stability to safe 3D trajectory selection and task placement.
- Attitude control: FEAR-PID for robust quadrotor attitude control across flight phases.
- Trajectory planning: ACS-DS with safety value and decoupling mechanism for reliable convergence.
- Task assignment: PSO-based decisions for where each offloaded task is executed.
- Resource allocation: jointly optimized with task assignment under cross-module feedback.
Figure 2. Holistic optimization pipeline overview.
High-Fidelity Simulators
To support fast iteration and showcase-quality visualization, we provide deployment-ready recipes across three simulators: a lightweight WebGL framework, Microsoft AirSim, and NVIDIA Isaac Sim.
Video 1. Web-based framework demo (fast iteration and sanity checks).
Video 2. Isaac Sim showcase (high-quality rendering).
Video 3. AirSim (AirSimNH Scene)
- Framework: reproducible planner/control visualization with lightweight runs.
- AirSim: high-fidelity scenes; supports two lines (pose-driven mainline and dynamics auxline).
- Isaac Sim: robust headless rendering pipeline; assets referenced by URL (no redistribution).
Key Results
We report extensive simulation results showing that the proposed holistic framework improves optimization convergence and delivers better end-to-end performance under realistic UAV-assisted fog computing settings.
The figures below highlight two representative outcomes: improved convergence behavior for trajectory planning, and reduced energy consumption when coupling control, mobility, and computing decisions.
- Improved convergence stability compared to classical baselines.
- Reduced energy consumption under comparable settings.
Figure 3. Convergence comparison (optimization stability and speed).
Figure 4. Energy consumption comparison.
Demos on Other Scenes
Figure 5. Web-based demo for Trajectory Planning (side-by-side).
Figure 6. Web-based demo for Attitude Control (side-by-side).
Video 4. AirSim (AbandonedPark Scene).
Video 5. AirSim (LandscapeMountains Scene).
Video 6. AirSim (Blocks Scene).
BibTeX
@article{LIU2026112064,
title={Energy-aware holistic optimization in UAV-assisted fog computing: Attitude, trajectory, and task assignment},
journal={Computer Networks},
volume={277},
pages={112064},
year={2026},
issn={1389-1286},
doi={https://doi.org/10.1016/j.comnet.2026.112064},
url={https://www.sciencedirect.com/science/article/pii/S1389128626000769},
author={Shuaijun Liu and Jinqiu Du and Yaxin Zheng and Jiaying Yin and Yuhui Deng and Jingjin Wu},
}