Software Defined Wide-Area Networks Control Plane Scaling Using Bayesian Game Model

Response time for control requests in Wide-Area Software-Defined (WA-SDN) networks crucially depends on processing delay and Wide-Area Networks (WAN) link bandwidth. During an overload condition, WAN OpenFlow (OF) routers are migrated from one controller to another, which causes Quality of Service (QoS) degradation for end-users. As the underlying topology is more sensitive to WAN delay, solving this problem becomes highly challenging from a Service Provider's perspective. This work introduces a Bayesian game framework for optimizing the OF routers-to-controller mapping in WAN. Further, the game theoretic optimizations lead to improved end-users utility, by optimal game module formulation. In the game model, each player achieves Bayesian Nash Equilibrium where the player decision implicitly takes care of WAN link bandwidth. The results obtained from emulating the proposed work, show better results in terms of QoS parameters, scalability and end-user utility.