Energy Efficient and Resilient Internet of Things Networks

Advancement in Internet-of-Things (IoT), mobile technologies and cloud computing services have inspired numerous designs for cloud-based real-time health monitoring systems. However, the massive transfer of health-related data to cloud contributes to increase the congestion in the networking infrastructure which leads to high latency and increased power consumption. Therefore, fog computing is introduced to provide service provisioning close to users. Nevertheless, the energy consumption of both transport network and processing infrastructures have yet to be probed further. Hence, this study proposes a new fog computing architecture under Gigabit Passive Optical Network (GPON) access network for health monitoring applications. A Mixed integer linear programming (MILP) model is introduced to optimise the number and locations of the processing servers at the network edge for energy-efficient fog computing. The model is developed for GPON and Ethernet access networks used to support fog processing. The impact of equipment idle power and the traffic volume have been investigated, and their effect on energy efficiency to serve low and high data rate health monitoring applications is established. The work also proposes resilient fog processing architectures for health monitoring applications. A MILP model for energy-efficient and resilient fog computing infrastructure considering two types of server protections related to geographic locations of primary and secondary processing servers are developed to optimise the number and locations of the processing servers at the network edge. In addition, a MILP model is developed to optimise energy efficiency and resilience of the proposed fog processing architectures considering server protection with geographical constraints and network protection with link and node disjoint resilience. The impact of increasing the level of resilience on the energy consumption of networking and processing is studied in contexts where the goal is to serve low and high data rate health monitoring applications.