Abstract: "The next era of the Internet of Things (IoT) calls for a large-scale deployment of edge devices to meet the growing demands of applications such as smart cities, smart grids, and environmental monitoring. From low-power sensors to multi-core platforms, IoT devices are prone to failures due to the reliability degradation of electronic circuits, batteries, and other components. As the network of heterogeneous devices expands, maintenance costs due to system failures become unmanageable, making reliability a major concern. Prior work has shown the importance of automated reliability management for meeting lifetime goals for individual devices. However, state-of-the-art network simulators do not provide reliability modeling capabilities for IoT networks. In this paper, we present an integrated reliability framework for IoT networks based on the ns-3 simulator. The lack of such tools restrained researchers from doing reliability-oriented analysis, exploration, and predictions early in the design cycle. Our contribution facilitates this, which can lead to the design of new network reliability management strategies. The proposed framework, besides reliability, incorporates three other interrelated models - power, performance, and temperature - which are required to model reliability. We validate our framework on a mesh network with ten heterogeneous devices, of three different types. We demonstrate that the models accurately capture the power, temperature, and reliability dynamics of real networks. We finally simulate and analyze two examples of energy-optimized and reliability-optimized network configurations to show how the framework offers an opportunity for researchers to explore trade-offs between energy and reliability in IoT networks."
Authors: Kazim Ergun, Xiaofan Yu and Nitish Nagesh (University of California San Diego, USA); Ludmila Cherkasova (ARM Research, USA); Pietro Mercati and Raid Ayoub (Intel Corporation, USA); Tajana Simunic Rosing (University of California, San Diego, USA)
Email: kergun@ucsd.edu, x1yu@ucsd.edu, nitish.nagesh@tum.de, lucy.cherkasova@gmail.com, pietro.mercati@intel.com, raid.ayoub@intel.com, tajana@cs.ucsd.edu