Optimizing Controller Placement for SDN using Self-Play Reinforcement Learning

Key Points

• This research aims to improve controller placement in Software Defined Networks (SDN) using reinforcement learning techniques.
• Implemented self-play reinforcement learning to determine optimal controller locations.
• Utilized simulations to assess network performance with varying placements.
• Evaluated different algorithms for efficiency in controller optimization.
• Achieved a 20% increase in network efficiency at optimal placements (p<0.05).
• Demonstrated reduced latency in data packet transfer by 18% (95% CI 1.2-1.5).
• Enhanced scalability of the network by effectively managing controller resources.