In recent years, the number of connected devices and systems has increased exponentially, along with their daily use. Cyber-Physical Systems (CPS) are complex multi-layered feedback systems that combine computing resources and interaction with the physical environment using sensors and actuators. Energy is considered one of the major concerns driven by the increasing number of connected systems. In this paper, we review current research approaches in energy-aware CPS. We propose a novel architectural methodology to analyse and compare state-of-the-art approaches, on their architectural design and energy-related factors. We finally draw recommendations from our review on how to build energy-aware CPS solutions.
16 November 2022 in IEEE Transactions on Sustainable Computing
The rise of the energy impact of software systems requires the need to optimize and reduce their energy consumption. One area often neglected is the important role played by users to drive energy reductions. In this paper, we aim to reduce the energy impact of software by pushing end users to change their software usage behavior, through raising awareness and providing software green feedback. We present a comprehensive and detailed field study of the impact of green feedback on software usage by end users, and the efficiency of green feedback on software behavioral change, using a distributed architecture aimed at providing accurate green feedback in real time. We find that green feedback helps in raising awareness about software energy, and on the willingness of users to apply energy-efficient changes. However, we also find that users lack the knowledge and tools to properly adopt lasting and energy-effective behavioral changes.
28 December 2022 in Sustainable Computing: Informatics and Systems
Computing demand in cloud environments has grown exponentially over the past decade, due to the increase in cloud workload related to new services such as artificial intelligence, autonomous vehicles, augmented reality, etc. As a result, the ICT sector has seen its carbon emissions increase. It is possible to adopt less energy-intensive strategies and consume electricity produced by renewable energy to limit the increase in carbon emissions. In this article, we present a review of the workload-shifting techniques available for sustainable workload deployment, providing an innovative framework that can be used to analyze energy-aware approaches that apply any type of shifting technique. We identified three main concepts: compute a workload at a different time, deploy a workload and/or its data in a different location, or use alternative processing to provide a good-enough option for a workload. A definition and some examples are given for each shifting concept, and then we explore the opportunities and challenges of combining different shifting techniques.