infrastructure sustainability planning

This study addresses the lack of generalized, systematic knowledge integration for Systems of Systems (SoS) in environmental sustainability. It conducts the first structured mapping study (SMS) encompassing the full spectrum of SoS—not merely subcategories such as smart cities. Leveraging automated searches across four academic databases and rigorous screening, the study identifies 39 high-quality publications from an initial pool of 926. It systematically analyzes SoS applications across carbon reduction, energy efficiency, and biodiversity conservation; identifies cross-cutting challenges—including interoperability and data governance; and reveals critical research gaps. Findings indicate dominance by smart cities and smart grids, while emerging domains—such as sustainable agriculture and wildfire management—remain severely underexplored. Based on this analysis, the study proposes six targeted future research directions, offering both a theoretical framework and practical guidance for advancing SoS-enabled environmental sustainability.

研究通过自下而上的举措，制定并实施可持续性策略，以帮助科学计算部门的数字研究基础设施转型，符合英国净零目标。

Current software design lacks systematic methods to assess conflicts between environmental requirements and quality attributes, as well as their broader sustainability implications. This paper proposes the Sustainability Impact Scoring (SIS) method at the architectural level: it is the first approach to systematically map environmental, social, and economic sustainability impacts onto the software quality attribute trade-off space; introduces a dynamic priority-weighting mechanism grounded in risk and stakeholder importance to enable quantified, cross-dimensional comparability of sustainability impacts; and integrates quality attribute analysis, risk-driven modeling, and a multi-model decision-making framework. Evaluated in an industrial energy-sector case study, SIS effectively identifies latent sustainability impacts of technical quality decisions, supports early preventive governance, enhances regulatory compliance, and strengthens strategic alignment with sustainability objectives.

This study identifies systematic value biases and institutional responsibility attribution biases in large language models (LLMs) regarding sustainability and AI’s societal role, aiming to inform organizational model selection under governance frameworks. Methodologically, it pioneers the use of standardized psychometric questionnaires—administered 100 times per model across five major LLMs (Claude, DeepSeek, GPT, LLaMA, Mistral)—combined with SDG-aligned quantitative evaluation and cross-model statistical comparison. Results reveal pronounced ideological divergence: GPT exhibits skepticism toward technological solutions for sustainability, whereas LLaMA displays extreme techno-optimism. The study empirically confirms that model choice significantly shapes sustainability-related decision outputs. It further identifies two governance-critical dimensions: “AI–sustainability compatibility cognition” and “responsibility attribution locus,” establishing an evidence-based foundation and methodological innovation for AI sustainability governance.

Existing software sustainability assessments predominantly focus on design and implementation phases, lacking actionable, scalable methods for the requirements engineering phase and relying heavily on manual effort, thereby impeding practical adoption. Method: This paper introduces the first sustainability framework tailored to requirements engineering, comprising three stages—identification, assessment, and optimization—and advances sustainability evaluation to the earliest requirements elicitation stage. It innovatively integrates a large language model (Gemini 2.5) with agent-based Retrieval-Augmented Generation (RAG), enabling semantic-driven, automated identification and optimization of sustainability requirements grounded in a structured sustainability requirement taxonomy. Contribution/Results: Evaluated across four cross-domain industrial projects, the framework significantly improves sustainability coverage at the requirements stage, overcomes the implementation gap of high-level sustainability guidelines, and establishes a novel paradigm for embedding environmental, social, technical, and economic sustainability objectives early in the software lifecycle.

Existing decision graph approaches struggle to effectively model safety and security requirements in adaptive systems. This work proposes an extended decision graph modeling language that, for the first time, incorporates a safety-event dimension within a sustainability-driven framework and enables unified, synergistic modeling of safety, security, and sustainability through multi-granular “safety modes.” The approach supports formal specification and divide-and-conquer fine-grained management of relevant scenarios across the entire system lifecycle. Experimental evaluation on an industrial collaboration use case demonstrates that the proposed extension more accurately captures complex safety and security scenarios, significantly enhancing the overall modeling capability for adaptive systems.

This work addresses the challenges in effectively evaluating and governing the sustainability of Green ICT systems, which stem from their structural complexity, fragmented metrics, and ambiguous accountability boundaries. To tackle these issues, the study proposes and preliminarily constructs a unified, cross-dimensional and multi-level reference framework that integrates perspectives from system architecture, life-cycle analysis, and diverse governance contexts. Developed by the Informatics Europe Green ICT Task Force, this framework transcends the limitations of single-metric approaches and lays the groundwork for standardization and practical implementation of Green ICT initiatives.

This study addresses the fragmented understanding of environmental impacts across the full lifecycle of artificial intelligence systems, a gap marked by incomplete phase coverage, inconsistent metrics, and opaque methodologies in current “green AI” research. The authors propose a unified analytical framework encompassing eight stages—from hardware manufacturing and data processing to model training and deployment—and integrate life cycle assessment (LCA), systematic literature review, and multidimensional environmental indicators such as CO₂e emissions. Their analysis reveals that existing studies predominantly focus on training and inference while overlooking critical factors like water consumption, raw material extraction, and embodied carbon. To rectify these omissions, the paper advances a standardized, comparable, and policy-oriented pathway for assessing AI’s environmental footprint, thereby promoting more systematic and rigorous green AI research.

This work addresses the unsustainable linear scaling paradigm of current generative AI systems, which neglects thermodynamic and material constraints, thereby exacerbating Scope 3 emissions and electronic waste and impeding the synergistic transition toward both digitalization and environmental sustainability. The study proposes a regenerative socio-technical roadmap that reconceptualizes AI infrastructure as a “system of systems” bounded by planetary limits. Integrating metabolic circuit frameworks, IEEE IRDS semiconductor sustainability guidelines, and circular economy principles, this approach centers on “value and need” rather than Nvidia-style centralized architectures. Emphasizing resource moderation and planetary responsibility, the proposed pathway identifies critical gaps in existing roadmaps and offers an actionable framework for digital circular economies that supports regulatory compliance and industrial resilience.

This study addresses the widespread lack of rigor in model construction, calibration, and integration within current ICT life cycle assessment (LCA) practices, which undermines result credibility. To tackle this issue, the work proposes four foundational trustworthiness criteria for ICT LCA—model provenance, clearly defined scope, end-to-end traceability, and obsolescence management—and introduces an open computational LCA framework grounded in these principles. The framework integrates dependency graph modeling, a versioned model repository, automated enforcement of completeness constraints, and a structured model taxonomy. It effectively uncovers common misuses and structural flaws in existing LCA studies, thereby offering a viable pathway toward transparent, reproducible, and sustainably evolvable assessments of ICT environmental impacts.