Enhancing Performance of Wide Area CIoT SDN by US-ML Based Optimum Controller Placement

Abstract

It is a critical area of study for enhancing the effectiveness of wide-area Cellular Internet of Things (CIoT) networks. One solution is to merge Software Defined Networking (SDN) with Internet of Things (IoT) network to boost efficiency. The main challenge is determining the best location for the SDN controller and evaluating SDN clustering. This paper proposed an Un-Supervised Machine-Learning (US-ML) approach based on silhouette distance along with gap statistic for finding the optimum number of controllers for network under consideration. In addition, the Partition Around Medoids (PAM) approach is opted for allocation of controller locations. Apart from SDN, another approach is to create efficient Low-Power Wide Area Networks (LPWAN). As a result, this research contributed to the study of various LPWAN design approaches and offered a method of optimal controller location for IoT-SDN cellular networks in industries. Several outstanding research challenges are noted, and prospective research objectives for LPWAN are offered. For the case study of wide area networks (WAN), a graphical representation of the SDN controller positioning method is presented. It is determined that effective placement can improve SDN performance in worst-case network scenarios.