Stochastic modelling of SDN controller for Internet of Vehicles

Internet of Vehicles (IoV) cannot be furnished with advanced networking features using standard network parameters like Quality of Service (QoS) and bandwidth management. To meet the QoS and bandwidth requirements for the vehicles, SDN controllers are used for provisioning. In this paper, the main objectives are to run the switching logic on the edge and core switches at the control plane, to balance the network bandwidth load and provisioning the bandwidth requirements to the Internet of Vehicles. State transition modelling and selecting optimal policies at SDN controllers are implemented in this work to better handle the bandwidth and to limit connectivity failures. The best state transitions were projected as optimal learning policies at SDN controller. In this paper, a stochastic method is also considered to estimate bandwidth consumption for IoV. The SDN controllers were modelled using the Semi Markov Decision Process (SMDP) which was then solved using a reinforcement learning model which achieves the higher learning factor which efficiently handles the SDN controller's bandwidth through a variety of rules. The entire network is emulated using Mininet and RYU controllers. The states of the SDN controller is modelled in such a way that the bandwidth provisioning is handled better in our work. When compared to the controllers existing policies, our model improves the bandwidth usage by a factor of at least 29%.