An hybrid and proactive architecture based on SDN for Internet of Things

In this paper, we focus on network control functionalities and resources management in Internet of Things (IoT) using Software Defined Network (SDN) and Fog computing paradigms. We propose a new hybrid and programmable architecture to make an appropriate decision related to the control process (centralized or semi-distribute) according to different heterogeneous parameters. The centralized scenario is a classical approach of SDN. However, the semi-distributed scenario consists in introducing different level of control: principal controller, secondary controllers, and local controllers nodes. In order to adapt the architecture to network topology, we use Connected Dominating Sets (CDS) algorithm to select a relevant subset of nodes able to act as local controllers. The selection strategy is proposed using fuzzy set approach based on different network parameters: connectivity degree, link quality, and the distance in terms of hop number from the gateway. The obtained results show that the selection strategy gives good results in terms of local controller selection.