As digital services are increasingly being deployed and used in a variety of domains, the environmental impact of Information and Communication Technologies (ICTs) is a matter of concern. Artificial intelligence is driving some of this growth but its environmental cost remains scarcely studied. A recent trend in large-scale generative models such as ChatGPT has especially drawn attention since their training requires intensive use of a massive number of specialized computing resources. The inference of those models is made accessible on the web as services, and using them additionally mobilizes end-user terminals, networks, and data centers. Therefore, those services contribute to global warming, worsen metal scarcity, and increase energy consumption. This work proposes an LCA-based methodology for a multi-criteria evaluation of the environmental impact of generative AI services, considering embodied and usage costs of all the resources required for training models, inferring from them, and hosting them online. We illustrate our methodology with Stable Diffusion as a service, an open-source text-to-image generative deep-learning model accessible online. This use case is based on an experimental observation of Stable Diffusion training and inference energy consumption. Through a sensitivity analysis, various scenarios estimating the influence of usage intensity on the impact sources are explored.

EcoIndex has been proposed to evaluate the absolute environmental performance of a given URL using a score ranging from 0 to 100 (the higher, the better). In this article, we make a critical analysis of the initial approach and propose alternatives that no longer calculate a plain score but allow the query to be situated among other queries. The generalized critiques come with statistics and rely on extensive experiments (first contribution). Then, we move on to low-cost Machine Learning (ML) approaches (second contribution) and a transition before obtaining our final results (third contribution). Our research aims to extend the initial idea of analytical computation, i.e., a relation between three variables, in the direction of algorithmic ML computations. The fourth contribution corresponds to a discussion on our implementation, available on a GitHub repository. Along with the paper, we invite the reader to examine the question: What attributes make sense for our problem?, or equivalently, what is a relevant data policy for studying digital environmental impacts? Beyond computational questions, it is important for the scientific community to focus on this question in particular. We currently promote using well-established ML techniques because of their potential, which we discuss in the paper. However, we also question techniques for their frugality or otherwise. Our data science project is still at the data exploration stage. We also want to encourage synergy between technical expertise and business knowledge because this is fundamental for advancing the data project.

The global energy demand for digital activities is constantly growing. Computing nodes and cloud services are at the heart of these activities. Understanding their energy consumption is an important step towards reducing it. On one hand, physical power meters are very accurate in measuring energy but they are expensive, difficult to deploy on a large scale, and are not able to provide measurements at the service level. On the other hand, power models and vendor-specific internal interfaces are already available or can be implemented on existing systems. Plenty of tools, called software-based power meters, have been developed around the concepts of power models and internal interfaces, in order to report the power consumption at levels ranging from the whole computing node to applications and services. However, we have found that it can be difficult to choose the right tool for a specific need. In this work, we qualitatively and experimentally compare several software-based power meters able to deal with CPU or GPU-based infrastructures. For this purpose, we evaluate them against high-precision physical power meters while executing various intensive workloads. We extend this empirical study to highlight the strengths and limitations of each software-based power meter.

Pedestrians can display dangerous behaviors in urban environments. Crossing on a red lights is the most common of these. Some people do so intentionally, knowing the risks, while others do so unintentionally by following other pedestrians. To decrease the former behavior, a visual device known as a nudge (a sign depicting eyes as a form of social control) has been tested on three sites in Strasbourg. Neutral “child’s eyes” and “woman’s eyes” signs from a server were used as experimental conditions and compared to two control conditions, one with and one without a sign that depicted flowers. Individual, social, and environmental variables were recorded for 2967 pedestrians arriving at a red light, using video analysis. Generalized linear models were then realized to test the effect of eye images. The different signs did not have any effect or opposite effect of those desired results. Neither of the two eyes signs had a significant effect on the rate of red-light crossings. However, the sign for “child’s eyes” tended to reduce waiting time at the pedestrian crossing. The results also show that the “flower” control sign significantly increased pedestrian risk-taking behaviors when crossing at a red light. Gender, age, the presence of distractors or other pedestrians, and the site itself is said to influence pedestrian behavior. Although previous studies have shown the effectiveness of nudges with eyes signals, these images do not appear to have a desired beneficial effect on pedestrian crossing behavior.

OBJECTIVE:The Pedestrian Behaviour Scale (PBS) is a self-report questionnaire that distinguishes five dimensions of pedestrian behaviour: violations, errors, lapses, aggressive behaviours and positive behaviours. This study aimed to meet three objectives: to trace the development of the PBS worldwide from 1997 to 2021, to report on its varied uses and to analyse the scientific validation of the different dimensions of pedestrian behaviour reflected by the PBS and its derivatives.DESIGN/METHODOLOGY:In this systematic literature review, we selected all works that cited the 2013 founding study of the PBS as well as all publications that cited the 2017 US validation of PBS which was frequently replicated around the world. We conducted an online database search using Web of Science, Google Scholar, ResearchGate and PubMed. After excluding duplicates, 116 studies were identified. A total of 30 studies were selected to meet our first two objectives and 14 studies were selected to meet our third objective.RESULTS:Over time, the PBS has undergone many changes. Overall, we found differences in the scientific validation of this questionnaire depending on the version used, the validation tests performed and the population studied. The original version of the PBS and its Turkish adaptation proved most appropriate for assessing the transgressions dimension. The American version of the PBS proved a suitable alternative but it is more suited to assessing the two independent dimensions of violations and errors. The Chinese version of the PBS (CPBS) proved unsuitable for assessing the lapses dimension, while the original version of the PBS emerged as the best option for assessing aggressive behaviours. The positive behaviour dimension presented many validation difficulties but its assessment by the CPBS seems to be the most appropriate option.CONCLUSION:As no systematic review of the PBS has been conducted before, researchers can now make an informed choice of methodology quickly and be guided by our recommendations regarding the use and possible improvements of the different validated versions.

Getting rid of the Coordinator in the IOTA Tangle is a challenging task, especially regarding the network trustworthiness. Using a customised testbed, we experimentally analyse the functioning of the GoShimmer, IOTA’s current implementation of the decentralised Tangle, with respect to specific network performance metrics. We observe that a trade-off exists among such metrics. We thus propose to determine the optimal rate allocation through an optimization problem maximising network performance and user utility. We further propose a distributed and asynchronous scheme to allow nodes to solve such problem.

Road accidents involving pedestrians are a reality of urban life. Pedestrian risk is now well known and documented from the perspective of drivers. However, pedestrian behaviour plays a central role in road accidents, notably in terms of illegal road crossing at signalized intersections. This study focuses on pedestrians crossing illegally at a signal light, and specifically investigates uncertainty behaviour, also referred to as hesitation, which occurs when a pedestrian slows down or stops his/her crossing movement then (1) abandons the crossing by returning to the kerb or (2) accelerates to cross the road more quickly. We sought to understand the causes of this behaviour in France and Japan, two countries where interesting differences have already been demonstrated in the way pedestrians behave. The results show a longer period of uncertainty for pedestrians in Japan compared to France. Japanese pedestrians also hesitated longer when they were alone. This study demonstrates a tendency to speed up if there are a number of pedestrians already crossing the road, but abandoning behaviours were more frequently observed than acceleration. This study confirms that pedestrians may misevaluate the moment to cross and hesitate when they realise that they have made a mistake, thus increasing the risk of an accident. These results could help to find solutions that prevent illegal and dangerous road-crossing behaviours.