Reproduce “Multi-Agent DRL for Queue-Aware Task Offloading in Hierarchical MEC-Enabled Air-Ground Networks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Muhammet Hevesli, Abegaz Mohammed Seid, Aiman Erbad, Mohamed Abdallah
Researchers developed MAPPO-BD, a Heterogeneous Multi-Agent Deep Reinforcement Learning algorithm, to jointly optimize UAV trajectories, queue-aware task offloading, and resource allocation in hierarchical MEC-enabled air-ground networks. This approach achieved up to 32% greater energy savings compared to baselines and provided higher hotspot fairness by efficiently managing computational demands and task queues.
Key insights
- Optimizing UAV trajectories is critical for efficient load distribution and minimizing energy consumption in hierarchical air-ground networks.
- Utilizing a Beta distribution in Deep Reinforcement Learning for heterogeneous agents with bounded action spaces leads to more effective policy learning and uniform exploration by avoiding action clipping.
- Distributing tasks among multiple UAVs significantly improves energy efficiency and scalability by better distributing the workload and reducing reliance on the High Altitude Platform Station (HAPS).
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.