2018/06/25 by Alexios Lekidis, Panagiotis Katsaros · 11 citations
Computer Science · Engineering · #Battery (electricity) #Component (thermodynamics) #Efficient energy use #Energy (signal processing) #Energy consumption #Green IT and Sustainability #Internet of Things #IoT and Edge/Fog Computing #Representation (politics) #Software System Performance and Reliability #cs.NI #eess.SP
paper · pdf · doi:10.4204/eptcs.272.3
published in Electronic Proceedings in Theoretical Computer Science 272, 24-38 (Open Publishing Association) · In Proceedings MeTRiD 2018, arXiv:1806.09330
openalex publication_date 2018/06/25 · arxiv created 2018/06/26 · arxiv updated 2018/07/04 · openalex created_date 2018/07/10 · openalex updated_date 2026/08/05
A major challenge that is currently faced in the design of applications for the Internet of Things (IoT) concerns with the optimal use of available energy resources given the battery lifetime of the IoT devices. The challenge is derived from the heterogeneity of the devices, in terms of their hardware and the provided functionalities (e.g data processing/communication). In this paper, we propose a novel method for (i) characterizing the parameters that influence energy consumption and (ii) validating the energy consumption of IoT devices against the system's energy-efficiency requirements (e.g. lifetime). Our approach is based on energy-aware models of the IoT application's design in the BIP (Behavior, Interaction, Priority) component framework. This allows for a detailed formal representation of the system's behavior and its subsequent validation, thus providing feedback for enhancements in the pre-deployment or pre-production stages. We illustrate our approach through a Building Management System, using well-known IoT devices running the Contiki OS that communicate by diverse IoT protocols (e.g. CoAP, MQTT). The results allow to derive tight bounds for the energy consumption in various device functionalities, as well as to validate lifetime requirements through Statistical Model Checking.