Energy Efficiency in Large-scale Internet of Things Networks

Abstract

Near-ubiquitous wireless connectivity combined with advancements in hardware and battery technology have enabled a proliferation of Internet-connected consumer and industrial devices. From smartphones in users' hands to Internet of Things (IoT) nodes streaming data from factory equipment, these devices have enabled new categories of services that have become a vital part of the consumer and industrial markets. Although various hardware and software aspects of wireless devices have seen immense improvements, battery technology still remains on a linear improvement path and optimization of the physical layer of the network stack is giving diminishing returns. Thus, the need to optimize the energy usage of the whole network stack still remains. This thesis focuses on the challenges related to understanding the effect of network traffic transmissions on the energy usage of the device. Specifically, research questions posited in this thesis look at the short-term predictability of user-driven network traffic and whether it can be exploited for traffic scheduling. Also, we explore what other variables affect energy usage during network transmissions and how they can be used in activity and energy models, which are needed to create less intrusive, leaner and more scalable energy profiling tools. For our research work, we chose two example scenarios - Android smartphones and IoT motes on a IEEE 802.15.4 network - each representing one of our focus networks.