Paper Title
HYBRIDIZED CONTROLLER PLACEMENT MODEL IN SOFTWARE DEFINED NETWORKING

Abstract
Abstract - Software defined networking (SDN) was occasioned by the challenges of programmability, maintainability and manageability, due to network ossification. To decouple the control system from the infrastructure plane caused by ossification the problem of controller placement arises.There is need to effectively place controllers within a network topology to manage the network of data plane devices from the control plane. The study adopted the Objected-Oriented Analysis and Design (OOAD) methodology in the analysis of the existing systems and the design of the hybrid system. A hybridized algorithms for controller placement (ACPM) mechanism was developed, to address the challenges of controller placement. The study used the unified modelling language (UML) tools in the design of the hybridized mechanism, and implemented the mechanism with the C# programming language, which is a bundled component of Microsoft Visual Studio integrated development environment (MVSIDE). The metrics of measure of the scalability was based on latencies, which are categories into four different latency metrics; Average Case Latencies for Switch to Controller, Worst Case Latencies for Switch to Controller, Average Case Latencies for Switch to Controller and, Worst Case Latencies for Controller to Controller. The hybrid POCO-MOCO Controller performs better than POCO and MOCO individually in the four measurement scenarios, as the latencies improves with more controllers within the network, reducing the latencies; from 3100 ms and 2500 ms for POCO and MOCO respectively to 2013 ms for Hybrid POCO-MOCO in Switch to Controller Average Case latency, from 2598 ms and 2769 for POCO and MOCO respectively to 2209 ms for Hybrid POCO-MOCO in Worst Case Switch to Controller latency, 2776 ms and 2987 ms for POCO and MOCO respectively to 1968 ms in Average Case Controller to Controller latency, and 2984 ms and 2759 for POCO and MOCO respectively to 2163 ms in Worst Case Controller to Controller latency. The hybrid POCO-MOCO Controller executes better than POCO and MOCO individually, in the four measurement scenarios, as the latencies improves with more controllers added within the network that reduced the latencies; from 2473 ms and 2129 ms for POCO and MOCO respectively to 2098 ms for Hybrid POCO-MOCO in Switch to Controller Average Case latency, from 2198 ms and 2268 for POCO and MOCO respectively to 2124 ms for Hybrid POCO-MOCO in Worst Case Switch to Controller latency, 2598 ms and 2471 ms for POCO and MOCO respectively to 2299 ms in Average Case Controller to Controller latency, and 2689 ms and 2814 for POCO and MOCO respectively to 2365 ms in Worst Case Controller to Controller latency. The hybrid POCO-MOCO controller performs marginally better than POCO and MOCO individually, in the four measurement scenarios, as the latencies improves with more controllers added to the network that reduced the latencies; from 2912 ms and 2784 ms for POCO and MOCO respectively to 2539 ms for Hybrid POCO-MOCO in Switch to Controller Average Case latency, from 3129 ms and 3017 for POCO and MOCO respectively to 2773 ms for Hybrid POCO-MOCO in Worst Case Switch to Controller latency, 2789 ms and 2693 ms for POCO and MOCO respectively to 2598 ms in Average Case Controller to Controller latency, and 2873 ms and 2756 for POCO and MOCO respectively to 2614 ms in Worst Case Controller to Controller latency. Keywords - Controller Placement, Software Defined Network, Hybridized Controller Placement Mechanism, Network Ossification.