Managing a Resilient Multitier Architecture for Unstable IoT Networks in Location Based-Services
Abstract
Facilitated by the widespread adoption of Internet of Things (IoT) networks, Location-based services (LBS) have emerged as a new type of services, requiring a high quality of service (QoS) and to provide access to all devices within predefined zones of interest. This is made possible via specific IoT Networks architectures based on the Software Defined Network paradigm. To address the challenge of unstable IoT networks management, where devices can move, appear, or vanish unpredictably, we propose a novel architecture based on a selection process of dominant devices acting as gateways, ensuring continuity of service. We investigate two selection processes, respectively based on Connected Dominating Sets and Deep Q-Network techniques. The objective of this method is to optimize energy consumption while providing high QoS and extending network access to offline devices within predefined zones of interest. In order to evaluate the performance of the proposed architecture with different selection processes, we conducted experiments using emulation tools allowing communication mode demand generations. The metrics used were the proportion of dominant devices, the energy consumption savings, the quality of service and the network extension to offline devices. Ultimately, we present a recommendation concerning the selection process based on the needs of the system.