Title: Optimising controller placement for SDN using self-play reinforcement learning

Authors: Benoudifa Ouafae; Abderrahim Ait Wakrime; Redouane Benaini

Addresses: Department of Computer Science, Faculty of Sciences, Mohammed V University, Rabat, Morocco ' Department of Computer Science, Faculty of Sciences, Mohammed V University, Rabat, Morocco ' Department of Computer Science, Faculty of Sciences, Mohammed V University, Rabat, Morocco

Abstract: The placement of the software-defined networking (SDN) controller is a critical component of SDN architecture, influencing the network's performance, scalability, reliability, and overall efficiency. The MuZero agent architecture offers an innovative approach to optimise coherence and efficiency in SDN controller placement. Known for its autonomous learning through decision- making and planning, the MuZero agent adapts to the complex challenge of controller placement. This study applies the MuZero agent to simulate scenarios where controllers are strategically assigned to nodes in an SDN, aiming to maximise network coherence while adhering to predefined constraints. The results demonstrate the successful optimisation of controller placement, showcasing the effectiveness of the MuZero approach in enhancing SDN performance.

Keywords: artificial intelligence; SDN; software-defined networking; data sharing; CPP; controller placement problem; MuZero Algorithm; network coherence; reinforcement learning; autonomous learning.

DOI: 10.1504/IJAACS.2026.154855

International Journal of Autonomous and Adaptive Communications Systems, 2026 Vol.19 No.3, pp.295 - 320

Received: 24 Jul 2024
Accepted: 29 May 2025
Published online: 16 Jul 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article