An SDN enhanced load balancing mechanism for a multi-controller WiFi network
Sohaib Manzoor, Syed Muhammad Abrar Akber, Muhammad Ilyas Menhas, Muhammad Imran, Muhammad Sajid, Hassan Talal, Usama Samad · 2018
Software Defined Networking (SDN) is a promising network architecture constructed by a decoupled data plane and control plane. Deployment of key entities i.e. controllers and switches is vital for achieving high performance environment. In this paper, we concentrate on classification, load balancing in switches and controllers (local and global). User requests arrive from WiFi Access Points (APs) (i.e.) WiFi network support. The novel SDN architecture involves 2-metric classifier in an AP and dynamic 2-level load balancing scheme (D2LLB) is applied over the global controller. Three significant features of classified flows are extracted and applied on Analytical Hierarchical Process (AHP), to allocate to controllers. Global Controller manages information of local controllers in form of clusters. Clustering is done in accordance to their processing information which is periodically updated. Clusters are un-loaded clusters, overloaded cluster and idle clusters. Hereby load is balanced in two ways; one is by migrating controllers in overloaded condition and second is allocating higher prioritized flows to idle controllers. Finally, our extensive simulation performed using OMNeT++ simulator, shows better performance in delay and file transfer as compared to weighted round robin technique and switch based load balancing.