Distributed Feedback Optimization of Linear Multi-agent Systems
(2024) Symposium on Systems Theory in Data and Optimization — Location: Stuttgart, Germany (30.September.2024)
Files
SysDO_template.pdf
Open Access - Adobe PDF
- 430.98 KB
Details
- Authors
- Abstract
- Feedback optimization has evolved as an important control approach for optimizing dynamical systems, particularly in terms of steady-state performance objectives. Traditional feedback optimization methods are primarily reliant on centralized systems and controller structures, which suffer from scalability and privacy issues for large-scale applications. This extended abstract investigates a distributed feedback optimization framework in which each agent updates its local control state by averaging inputs from neighboring agents and performing a local negative gradient step. Assuming convexity and smoothness of the cost, we show that this control technique converges to a fixed point. Further, under the weaker assumption of restricted strong convexity of the cost function, we prove that the algorithm achieves linear convergence to a vicinity of the optimal point, with the neighborhood size contingent on the chosen stepsize. Simulation results support the theoretical findings.
- Affiliations
Citations
Mehrnoosh, A., & Bianchin, G. (2024). Distributed Feedback Optimization of Linear Multi-agent Systems. Symposium on Systems Theory in Data and Optimization, Stuttgart, Germany. https://hdl.handle.net/2078.5/270613