Online Learning and Iterative Pricing for Large-Scale Satellite Scheduling
A novel framework tackles distributed constraint optimization problems (DCOPs) specifically for large-scale decentralized satellite scheduling. This method reexamines the relationship between DCOPs and potential games, integrating contemporary online learning algorithms to discover equilibria. These algorithms demonstrate competitiveness against typical incomplete DCOP algorithms. The framework splits a DCOP into two interrelated subproblems: a high-level meta-DCOP focused on task allocation and separate local optimization tasks for scheduling. An innovative iterative pricing technique refines meta-level utilities based on insights from local optimizations. This research has been published on arXiv under ID 2607.25835.
Key facts
- Distributed constraint optimization problems (DCOPs) are used for distributed decision making under limited communication.
- Many real-world DCOP instances are too large to solve monolithically.
- The research revisits the connection between DCOPs and potential games.
- Online learning algorithms for equilibrium finding are adapted to DCOPs.
- The proposed algorithms are competitive with representative incomplete DCOP algorithms.
- A new decomposition framework separates a DCOP into a meta-DCOP for task allocation and local optimization for scheduling.
- An iterative pricing method couples the two levels by updating meta-level utilities using feedback from local optimizations.
- The application is large-scale decentralized satellite scheduling.
Entities
Institutions
- arXiv