Sustainable Computing-Informatics & Systems最新文献

Multi fault classification in electrical transmission lines using feature engineering based on autogluon framework 基于自胶子框架的特征工程在输电线路多故障分类中的应用

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-02-04 DOI: 10.1016/j.suscom.2026.101310

Merve Demirci

With the increasing electricity demand, power transmission lines continue to grow and become increasingly complex. Because even the smallest faults occurring on growing power lines can impact the grid, rapid fault detection, classification, and subsequent repair are crucial. In this study, a new Machine Learning approach based on the Autogluon framework is proposed for rapid fault detection and high-accuracy classification of transmission line faults. Three different methods are employed within the Autogluon framework, and the results are evaluated and compared using ROC analysis. The first method uses the original dataset containing 7861 data points obtained from the open-source Kaggle platform. The second approach adds statistical properties of current and voltage values (mean, standard deviation, and range) to the original dataset. The third method uses the SMOTE algorithm to generate synthetic data and increase the data size to address the imbalance in the number of fault classes in the dataset. The proposed method demonstrates the critical role of data processing and feature engineering in optimizing classification performance and achieving the most accurate diagnosis for multi-label fault classification. The numerical results compared with the literature show that the proposed method is promising for practical applications.

{"title":"Multi fault classification in electrical transmission lines using feature engineering based on autogluon framework","authors":"Merve Demirci","doi":"10.1016/j.suscom.2026.101310","DOIUrl":"10.1016/j.suscom.2026.101310","url":null,"abstract":"<div><div>With the increasing electricity demand, power transmission lines continue to grow and become increasingly complex. Because even the smallest faults occurring on growing power lines can impact the grid, rapid fault detection, classification, and subsequent repair are crucial. In this study, a new Machine Learning approach based on the Autogluon framework is proposed for rapid fault detection and high-accuracy classification of transmission line faults. Three different methods are employed within the Autogluon framework, and the results are evaluated and compared using ROC analysis. The first method uses the original dataset containing 7861 data points obtained from the open-source Kaggle platform. The second approach adds statistical properties of current and voltage values (mean, standard deviation, and range) to the original dataset. The third method uses the SMOTE algorithm to generate synthetic data and increase the data size to address the imbalance in the number of fault classes in the dataset. The proposed method demonstrates the critical role of data processing and feature engineering in optimizing classification performance and achieving the most accurate diagnosis for multi-label fault classification. The numerical results compared with the literature show that the proposed method is promising for practical applications.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"50 ","pages":"Article 101310"},"PeriodicalIF":5.7,"publicationDate":"2026-02-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146122686","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Day-ahead energy management in smart combined cooling, heating and power (CCHP) grid considering optimal consumption and local self-generation 考虑最优消耗和本地自产的智能冷热电联产电网日前能源管理

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2026.101291

Mohamad Reza Zargar Shoshtari, Seyed Mehdi Hakimi, Ghasem Derakhshan

With growing global energy demand, ensuring a reliable energy supply is critical for all nations. The modern Energy services in residential buildings, especially those using combined cooling, heating, and power (CCHP) systems, are particularly important in meeting these demands. Accordingly, this study focuses on day-ahead energy management in a smart CCHP grid with the participation of hybrid energy storage systems and optimal energy consumption by consumers in smart residential buildings. The energy management is modeled by a multi-level and multi-objective optimization approach considering demand response strategies (DRSs). The DRSs include electrical demand shifting of power consumption, and self-generation of power, and gas storage systems. The electrical demand shifting strategy is implemented in the first level optimization, subject to electricity pricing traffic to minimize consumers’ bills. Also, minimizing consumers’ bills in the second level optimization is done by power and gas storage systems via the local self-generation (LS-G) strategy, subject to electricity and gas prices in the energy market. In the third level optimization, multi-objective functions like minimizing operational costs, maximizing flexibility and minimizing power losses are implemented. In the proposed optimization approach, optimized energy consumption in the first and second levels is considered in the third level optimization. The proposed optimization approach for all levels is solved by using General Algebraic Modeling System (GAMS) software. In the following, solving multi-objective optimization approach in the third level is carried out by enhanced epsilon-constraint method. Also, Shannon Entropy decision making method is proposed for determining optimal solution in third level for multi-objective functions and Pareto front solutions. Finally, the findings show the optimal results of the objectives at each level and highlight consumer involvement through a comparative analysis via various case studies. The participation of DRSs leads to a 11.63 % reduction in operational costs and 18.75 % reduction in power losses, while also enhancing flexibility by 2.6 % in the CCHP grid.

随着全球能源需求的增长，确保可靠的能源供应对所有国家都至关重要。住宅建筑中的现代能源服务，特别是那些使用冷、热、电联产（CCHP）系统的能源服务，对于满足这些需求尤为重要。因此，本研究聚焦于混合储能系统参与的智能热电联产电网的日前能源管理，以及智能住宅建筑中消费者的最优能源消耗。采用考虑需求响应策略的多层次多目标优化方法对能源管理进行建模。DRSs包括电力消费的电力需求转移、自产电和储气系统。在第一级优化中实施电力需求转移策略，在电价流量的约束下实现用户电费最小化。此外，根据能源市场的电力和天然气价格，电力和天然气存储系统通过本地自我发电（LS-G）策略实现了第二级优化中消费者账单的最小化。在第三层优化中，实现了运营成本最小化、灵活性最大化和功率损耗最小化等多目标函数。在本文提出的优化方法中，在第三级优化中考虑了第一级和第二级的优化能耗。采用通用代数建模系统（GAMS）软件对所提出的各级优化方法进行求解。下面，通过增强的epsilon约束方法求解第三层次的多目标优化问题。针对多目标函数和Pareto前解的三阶最优解，提出了Shannon熵决策方法。最后，研究结果显示了每个层次目标的最佳结果，并通过各种案例研究的比较分析突出了消费者的参与。drs的参与使运行成本降低11.63% %，电力损耗降低18.75% %，同时也使CCHP电网的灵活性提高了2.6% %。

{"title":"Day-ahead energy management in smart combined cooling, heating and power (CCHP) grid considering optimal consumption and local self-generation","authors":"Mohamad Reza Zargar Shoshtari, Seyed Mehdi Hakimi, Ghasem Derakhshan","doi":"10.1016/j.suscom.2026.101291","DOIUrl":"10.1016/j.suscom.2026.101291","url":null,"abstract":"<div><div>With growing global energy demand, ensuring a reliable energy supply is critical for all nations. The modern Energy services in residential buildings, especially those using combined cooling, heating, and power (CCHP) systems, are particularly important in meeting these demands. Accordingly, this study focuses on day-ahead energy management in a smart CCHP grid with the participation of hybrid energy storage systems and optimal energy consumption by consumers in smart residential buildings. The energy management is modeled by a multi-level and multi-objective optimization approach considering demand response strategies (DRSs). The DRSs include electrical demand shifting of power consumption, and self-generation of power, and gas storage systems. The electrical demand shifting strategy is implemented in the first level optimization, subject to electricity pricing traffic to minimize consumers’ bills. Also, minimizing consumers’ bills in the second level optimization is done by power and gas storage systems via the local self-generation (LS-G) strategy, subject to electricity and gas prices in the energy market. In the third level optimization, multi-objective functions like minimizing operational costs, maximizing flexibility and minimizing power losses are implemented. In the proposed optimization approach, optimized energy consumption in the first and second levels is considered in the third level optimization. The proposed optimization approach for all levels is solved by using General Algebraic Modeling System (GAMS) software. In the following, solving multi-objective optimization approach in the third level is carried out by enhanced epsilon-constraint method. Also, Shannon Entropy decision making method is proposed for determining optimal solution in third level for multi-objective functions and Pareto front solutions. Finally, the findings show the optimal results of the objectives at each level and highlight consumer involvement through a comparative analysis via various case studies. The participation of DRSs leads to a 11.63 % reduction in operational costs and 18.75 % reduction in power losses, while also enhancing flexibility by 2.6 % in the CCHP grid.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101291"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Energy efficient unified computing framework for smart grids with AI-driven communication, supercomputing, and energy perception orchestration 智能电网的节能统一计算框架，具有人工智能驱动的通信、超级计算和能源感知编排

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101289

Wenchong Fang , Zhifeng Zhou , Yingchen Li , Ma Guang , Fei Chen

The next-generation smart grid requires a unified computing framework that seamlessly integrates communication, high-performance computing (HPC), and AI to enable real-time energy perception, forecasting, and decision-making. Conventional architectures, which treat communication, computation, and control as independent modules, often suffer from latency, scalability limitations, and weak coordination across heterogeneous infrastructures. To overcome these constraints, this work proposes an energy-efficient unified computing framework where communication networks, HPC clusters, and AI orchestration operate as a tightly coupled ecosystem. AI modules handle deep learning–based perception of multi-source energy data and employ reinforcement learning to optimize dynamic load allocation and demand-side flexibility. Superscale HPC resources accelerate renewable forecasting, grid stability assessment, and large-scale optimization tasks. In parallel, adaptive communication units with edge-level compression and intelligent routing ensure low latency and resilience under varying network loads. The framework is evaluated through MATLAB/Simulink and Python co-simulation using HPC-enabled TensorFlow clusters and blockchain-secured IoT gateways. Experimental results demonstrate a System Orchestration Index (SOI) of 98.3 %, a Computational Efficiency Ratio (CER) of 37.5 %, a Demand Flexibility Index (DFI) of 33.8 %, and an end-to-end decision latency of 18 ms. Compared with conventional grid computing approaches, the proposed architecture achieves improvements of 9.4 % in orchestration efficiency, 7.8 % in computational efficiency, and 6.2 % in demand flexibility. These outcomes highlight the potential of an AI-driven, HPC-accelerated, and communication-adaptive unified computing paradigm for scalable and resilient smart grid operations.

下一代智能电网需要一个统一的计算框架，将通信、高性能计算（HPC）和人工智能无缝集成，实现实时能源感知、预测和决策。将通信、计算和控制视为独立模块的传统体系结构经常受到延迟、可伸缩性限制和跨异构基础结构的弱协调的困扰。为了克服这些限制，本研究提出了一种节能的统一计算框架，其中通信网络、高性能计算集群和人工智能编排作为一个紧密耦合的生态系统运行。人工智能模块处理基于深度学习的多源能源数据感知，并采用强化学习来优化动态负载分配和需求侧灵活性。超大规模高性能计算资源加速可再生预测、电网稳定性评估和大规模优化任务。同时，具有边缘级压缩和智能路由的自适应通信单元确保了在不同网络负载下的低延迟和弹性。该框架通过MATLAB/Simulink和Python联合仿真进行评估，使用支持hpc的TensorFlow集群和区块链安全的物联网网关。实验结果表明，系统编排指数（SOI）为98.3% %，计算效率比（CER）为37.5% %，需求灵活性指数（DFI）为33.8% %，端到端决策延迟为18 ms。与传统的网格计算方法相比，该架构的编排效率提高了9.4% %，计算效率提高了7.8 %，需求灵活性提高了6.2% %。这些结果突出了ai驱动、hpc加速和通信自适应的统一计算范式在可扩展和弹性智能电网运营中的潜力。

{"title":"Energy efficient unified computing framework for smart grids with AI-driven communication, supercomputing, and energy perception orchestration","authors":"Wenchong Fang , Zhifeng Zhou , Yingchen Li , Ma Guang , Fei Chen","doi":"10.1016/j.suscom.2025.101289","DOIUrl":"10.1016/j.suscom.2025.101289","url":null,"abstract":"<div><div>The next-generation smart grid requires a unified computing framework that seamlessly integrates communication, high-performance computing (HPC), and AI to enable real-time energy perception, forecasting, and decision-making. Conventional architectures, which treat communication, computation, and control as independent modules, often suffer from latency, scalability limitations, and weak coordination across heterogeneous infrastructures. To overcome these constraints, this work proposes an energy-efficient unified computing framework where communication networks, HPC clusters, and AI orchestration operate as a tightly coupled ecosystem. AI modules handle deep learning–based perception of multi-source energy data and employ reinforcement learning to optimize dynamic load allocation and demand-side flexibility. Superscale HPC resources accelerate renewable forecasting, grid stability assessment, and large-scale optimization tasks. In parallel, adaptive communication units with edge-level compression and intelligent routing ensure low latency and resilience under varying network loads. The framework is evaluated through MATLAB/Simulink and Python co-simulation using HPC-enabled TensorFlow clusters and blockchain-secured IoT gateways. Experimental results demonstrate a System Orchestration Index (SOI) of 98.3 %, a Computational Efficiency Ratio (CER) of 37.5 %, a Demand Flexibility Index (DFI) of 33.8 %, and an end-to-end decision latency of 18 ms. Compared with conventional grid computing approaches, the proposed architecture achieves improvements of 9.4 % in orchestration efficiency, 7.8 % in computational efficiency, and 6.2 % in demand flexibility. These outcomes highlight the potential of an AI-driven, HPC-accelerated, and communication-adaptive unified computing paradigm for scalable and resilient smart grid operations.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101289"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938425","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Towards energy-efficient scientific computing: Reversible numerical linear algebra kernels in floating-point arithmetic 迈向节能科学计算：浮点运算中的可逆数值线性代数核

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101261

V. Dwarka

Frontier scientific and AI workloads now reach

1 0^{19} - 1 0^{25}

fused multiply–add (FMA) operations per run (on the order of

2 \times 1 0^{19} - 2 \times 1 0^{25}

FLOPs). At today’s

\sim 10

pJ per FMA, this corresponds to approximately

1 0^{8} - 1 0^{14}

joules of arithmetic energy. At this scale, energy becomes the limiting resource for continued growth in computational workloads, motivating a re-evaluation of long-standing algorithmic assumptions. It is often assumed that reversible computing only matters near the Landauer limit. Building on prior physical arguments that full energy recovery is only possible when computation preserves information, we demonstrate that this same requirement governs floating-point numerical kernels: overwriting state enforces a non-zero energy floor, even under ideal recovery. Thus, eliminating this wall in practice requires that the numerical algorithm itself be injective. We therefore present the first reversible floating-point realizations of core dense numerical kernels—matrix multiplication, LU factorization, and conjugate-gradient iteration—that retain rounding information rather than discarding it. Implemented directly in IEEE arithmetic, they achieve machine-precision forward–reverse agreement on well- and ill-conditioned problems with minimal auxiliary state. A toggle-based model with measured switching costs and realistic recovery factors predicts

1 0^{3} - 1 0^{4} \times

reductions in arithmetic energy. These results establish injective floating-point kernels as a foundation for energy-recovering numerical computation, and indicate that realizing this potential will require sustained co-design across applied mathematics, computer science, and hardware engineering.

前沿科学和人工智能工作负载现在达到每次运行1019−1025次融合乘加（FMA）运算（顺序为2×1019−2×1025 FLOPs）。在今天的~ 10 pJ / FMA下，这相当于大约108−1014焦耳的算术能量。在这种规模下，能源成为计算工作量持续增长的限制资源，促使人们对长期存在的算法假设进行重新评估。通常假设可逆计算只在兰道尔极限附近起作用。基于先前的物理论据，即只有在计算保留信息时才有可能完全恢复能量，我们证明了浮点数值核也有同样的要求：即使在理想的恢复情况下，覆盖状态也会强制实现非零能量底限。因此，在实践中消除这堵墙需要数值算法本身是内射的。因此，我们提出了核心密集数值核的第一个可逆浮点实现-矩阵乘法，LU分解和共轭梯度迭代-保留舍入信息而不是丢弃它。它们直接在IEEE算法中实现，以最小的辅助状态实现对良好和病态问题的机器精度的正反向一致。一个基于开关的模型与测量的开关成本和现实的恢复因子预测103 - 104倍的算术能量降低。这些结果确立了注入浮点核作为能量回收数值计算的基础，并表明实现这一潜力将需要应用数学、计算机科学和硬件工程之间持续的协同设计。

{"title":"Towards energy-efficient scientific computing: Reversible numerical linear algebra kernels in floating-point arithmetic","authors":"V. Dwarka","doi":"10.1016/j.suscom.2025.101261","DOIUrl":"10.1016/j.suscom.2025.101261","url":null,"abstract":"<div><div>Frontier scientific and AI workloads now reach <span><math><mrow><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>19</mn></mrow></msup><mspace></mspace><mo>−</mo><mspace></mspace><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>25</mn></mrow></msup></mrow></math></span> fused multiply–add (FMA) operations per run (on the order of <span><math><mrow><mn>2</mn><mo>×</mo><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>19</mn></mrow></msup><mspace></mspace><mo>−</mo><mspace></mspace><mn>2</mn><mo>×</mo><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>25</mn></mrow></msup></mrow></math></span> FLOPs). At today’s <span><math><mrow><mo>∼</mo><mn>10</mn></mrow></math></span> pJ per FMA, this corresponds to approximately <span><math><mrow><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>8</mn></mrow></msup><mspace></mspace><mo>−</mo><mspace></mspace><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>14</mn></mrow></msup></mrow></math></span> joules of arithmetic energy. At this scale, energy becomes the limiting resource for continued growth in computational workloads, motivating a re-evaluation of long-standing algorithmic assumptions. It is often assumed that reversible computing only matters near the Landauer limit. Building on prior physical arguments that full energy recovery is only possible when computation preserves information, we demonstrate that this same requirement governs floating-point numerical kernels: overwriting state enforces a non-zero energy floor, even under ideal recovery. Thus, eliminating this wall in practice requires that the numerical algorithm itself be injective. We therefore present the <em>first</em> reversible floating-point realizations of core dense numerical kernels—matrix multiplication, LU factorization, and conjugate-gradient iteration—that retain rounding information rather than discarding it. Implemented directly in IEEE arithmetic, they achieve machine-precision forward–reverse agreement on well- and ill-conditioned problems with minimal auxiliary state. A toggle-based model with measured switching costs and realistic recovery factors predicts <span><math><mrow><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>3</mn></mrow></msup><mspace></mspace><mo>−</mo><mspace></mspace><mn>1</mn><msup><mrow><mn>0</mn></mrow><mrow><mn>4</mn></mrow></msup><mo>×</mo></mrow></math></span> reductions in arithmetic energy. These results establish injective floating-point kernels as a foundation for energy-recovering numerical computation, and indicate that realizing this potential will require sustained co-design across applied mathematics, computer science, and hardware engineering.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101261"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884063","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Energy efficient optimization of peer-to-peer energy exchanges in microgrids using blockchain-enabled federated learning 使用支持区块链的联邦学习对微电网中点对点能源交换进行节能优化

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101290

Raenu Kolandaisamy , A. Arun Kumar , A.N. Sasikumar , V. Sivakumar , G.K. Kamalam , M.K. Mohamed Faizal , Muhammad Mohzary

The growth of the penetration of distributed renewable resources has enhanced the need to have secure, scalable, and energy-efficient P2P (peer-to-peer) energy trading in microgrid settings. Conventional centralized market systems and aggregator-based trading systems can be limited by privacy leakage, high costs of communication, and single points of failure, which restrict them in the scalability of decentralized environments. The proposed blockchain-enabling federated learning (FL) framework that is suggested in this paper will aim to optimise P2P energy exchanges in microgrids to mitigate the limitations of the existing approach. Smart meters based on the Internet-of-Things record data on generation and consumption, which is locally computed to learn federated models without revealing raw information about users. The FL module uses gradient sharing together with adaptive aggregation to optimize the exchange decisions among prosumers, and the blockchain component can guarantee trade settlement with tamper-proof and the contract enforcement through proof-of-stake consensus mechanism that is based on lightweight cryptography. Moreover, edge-level compression of data and incentive sensitive participation schemes minimize communication costs and encourage fair participation in the market. Experiments of performance measurements on a MATLAB/Simulink-TensorFlow co-simulation model with Hyperledger Fabric show trading efficiency of 95.7 % and 27.6 % energy savings improvement, throughput of 174 transactions per second and 31 % communication overhead reduction compared to centralized and non-blockchain FL schemes. The findings support the ability of the proposed system to provide privacy preserving, energy saving, and safe P2P energy transactions in next-generation decentralized microgrid ecosystems.

分布式可再生资源渗透的增长增强了对微电网环境中安全、可扩展和节能的P2P（点对点）能源交易的需求。传统的集中式市场系统和基于聚合器的交易系统可能受到隐私泄露、通信成本高和单点故障的限制，这些限制了它们在分散环境中的可扩展性。本文提出的支持区块链的联邦学习（FL）框架旨在优化微电网中的P2P能源交换，以减轻现有方法的局限性。基于物联网的智能电表记录发电量和用电量的数据，这些数据是在本地计算的，以学习联邦模型，而不会泄露用户的原始信息。FL模块使用梯度共享和自适应聚合来优化产消者之间的交易决策，区块链组件通过基于轻量级加密的权益证明共识机制来保证防篡改的贸易结算和合同执行。此外，边缘级数据压缩和激励敏感参与方案最大限度地降低了沟通成本，并鼓励公平参与市场。在具有Hyperledger Fabric的MATLAB/Simulink-TensorFlow联合仿真模型上进行的性能测量实验表明，与集中式和非区块链FL方案相比，交易效率提高了95.7% %，节能提高了27.6% %，每秒174笔交易的吞吐量减少了31% %的通信开销。研究结果支持了该系统在下一代分散式微电网生态系统中提供隐私保护、节能和安全P2P能源交易的能力。

{"title":"Energy efficient optimization of peer-to-peer energy exchanges in microgrids using blockchain-enabled federated learning","authors":"Raenu Kolandaisamy , A. Arun Kumar , A.N. Sasikumar , V. Sivakumar , G.K. Kamalam , M.K. Mohamed Faizal , Muhammad Mohzary","doi":"10.1016/j.suscom.2025.101290","DOIUrl":"10.1016/j.suscom.2025.101290","url":null,"abstract":"<div><div>The growth of the penetration of distributed renewable resources has enhanced the need to have secure, scalable, and energy-efficient P2P (peer-to-peer) energy trading in microgrid settings. Conventional centralized market systems and aggregator-based trading systems can be limited by privacy leakage, high costs of communication, and single points of failure, which restrict them in the scalability of decentralized environments. The proposed blockchain-enabling federated learning (FL) framework that is suggested in this paper will aim to optimise P2P energy exchanges in microgrids to mitigate the limitations of the existing approach. Smart meters based on the Internet-of-Things record data on generation and consumption, which is locally computed to learn federated models without revealing raw information about users. The FL module uses gradient sharing together with adaptive aggregation to optimize the exchange decisions among prosumers, and the blockchain component can guarantee trade settlement with tamper-proof and the contract enforcement through proof-of-stake consensus mechanism that is based on lightweight cryptography. Moreover, edge-level compression of data and incentive sensitive participation schemes minimize communication costs and encourage fair participation in the market. Experiments of performance measurements on a MATLAB/Simulink-TensorFlow co-simulation model with Hyperledger Fabric show trading efficiency of 95.7 % and 27.6 % energy savings improvement, throughput of 174 transactions per second and 31 % communication overhead reduction compared to centralized and non-blockchain FL schemes. The findings support the ability of the proposed system to provide privacy preserving, energy saving, and safe P2P energy transactions in next-generation decentralized microgrid ecosystems.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101290"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Balancing carbon footprint and algorithm performance in recommender systems: A comprehensive benchmark 平衡推荐系统中的碳足迹和算法性能：一个全面的基准

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101286

Giuseppe Spillo , Allegra De Filippo , Cataldo Musto , Michela Milano , Giovanni Semeraro

In this paper, we present a reproducible pipeline to benchmark the trade-off between carbon emissions and recommendation performance across 14 algorithms and three publicly available datasets. In particular, we contribute: (a) a standardized protocol to account for carbon emissions of recommendation algorithms; (b) an empirical quantification of the carbon cost of hyperparameter tuning, and (c) an evaluation of data-reduction strategies as a low-cost approach to reduce emissions while improving certain non-accuracy metrics. Unlike previous literature, which mainly focused on the trade-off between performance and emissions, our benchmark reveals the cost of hyperparameter tuning. It examines the impact of data reduction techniques on the path toward sustainability-aware recommender systems. Our results show that simpler algorithms often deliver competitive accuracy at significantly lower emissions, and that exhaustive tuning can dramatically increase carbon costs with limited accuracy gains. Generally speaking, this study aims to discuss the challenges of energy consumption in recommender systems and to develop a new generation of algorithms that prioritize sustainability. All code and experiment traces are publicly released for reproducibility on Github.¹

在本文中，我们提出了一个可重复的管道，用于在14种算法和三个公开可用的数据集上对碳排放和推荐性能之间的权衡进行基准测试。具体来说，我们贡献了：(a)一个标准化的协议来解释推荐算法的碳排放；(b)对超参数调整的碳成本进行实证量化，以及(c)评估数据缩减策略作为一种低成本方法，在改善某些非准确性指标的同时减少排放。与以前的文献不同，这些文献主要关注性能和排放之间的权衡，我们的基准揭示了超参数调优的成本。它研究了数据减少技术对实现可持续意识推荐系统的影响。我们的研究结果表明，更简单的算法通常在显著降低排放的情况下提供具有竞争力的精度，而详尽的调整可以在有限的精度增益下显着增加碳成本。总的来说，本研究旨在讨论推荐系统中能源消耗的挑战，并开发优先考虑可持续性的新一代算法。所有代码和实验痕迹都在Github.1上公开发布，以便再现

{"title":"Balancing carbon footprint and algorithm performance in recommender systems: A comprehensive benchmark","authors":"Giuseppe Spillo , Allegra De Filippo , Cataldo Musto , Michela Milano , Giovanni Semeraro","doi":"10.1016/j.suscom.2025.101286","DOIUrl":"10.1016/j.suscom.2025.101286","url":null,"abstract":"<div><div>In this paper, we present a reproducible pipeline to benchmark the trade-off between carbon emissions and recommendation performance across 14 algorithms and three publicly available datasets. In particular, we contribute: <em>(a)</em> a standardized protocol to account for carbon emissions of recommendation algorithms; <em>(b)</em> an empirical quantification of the carbon cost of hyperparameter tuning, and <em>(c)</em> an evaluation of data-reduction strategies as a low-cost approach to reduce emissions while improving certain non-accuracy metrics. Unlike previous literature, which mainly focused on the trade-off between performance and emissions, our benchmark reveals the cost of hyperparameter tuning. It examines the impact of data reduction techniques on the path toward sustainability-aware recommender systems. Our results show that simpler algorithms often deliver competitive accuracy at significantly lower emissions, and that exhaustive tuning can dramatically increase carbon costs with limited accuracy gains. Generally speaking, this study aims to discuss the challenges of energy consumption in recommender systems and to develop a new generation of algorithms that prioritize sustainability. All code and experiment traces are publicly released for reproducibility on Github.<span><span><sup>1</sup></span></span></div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101286"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

AquaSense-AMC: An adaptive modulation-control model for energy-efficient communication in underwater IoT networks AquaSense-AMC：水下物联网网络中节能通信的自适应调制控制模型

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101263

S.R. Janani , Poornima. S , Aruna R , Chandru Vignesh C

The Internet of Underwater Things (IoUT) is revolutionizing maritime research, environmental monitoring and intelligent aquatic applications with real-time sensing and communication. Energy-efficient communication is the biggest challenge because of the constraint of bandwidth, long delay of propagation and highly dynamic water channels. To combat these issues, AquaSense-AMC is introduced a new Adaptive Modulation-Control (AMC) Model for underwater IoT networks. The framework provides three novel components: Channel-Aware Modulation Switching (CAMS), dynamically switching between modulation depth and symbol rate based on channel fluctuations; Energy-Constrained Control Mechanism (ECCM), energy-optimizing transmission power with predictive energy management to maximize node lifetime; and Hybrid Acoustic-Optical Relay (HAOR), a two-mode relay scheme that integrates acoustic links for extreme distance reliability with optical links for high-rate near-distance data transfer. Experimental assessments prove that AquaSense-AMC saves energy by 28 %, enhances packet delivery ratio by 35 % and increases network lifetime by 22 % relative to baseline methods. The model implements a sustainable and adaptive communication system and making IoUT operation reliable and energy-efficient in intricate underwater environments.

水下物联网（IoUT）正在通过实时传感和通信彻底改变海洋研究、环境监测和智能水生应用。由于带宽的限制、传播的长延迟和高度动态的水通道，节能通信是最大的挑战。为了解决这些问题，AquaSense-AMC为水下物联网网络引入了一种新的自适应调制控制（AMC）模型。该框架提供了三个新组件：信道感知调制开关（CAMS），基于信道波动在调制深度和符号率之间动态切换；能量约束控制机制（ECCM），具有预测能量管理的能量优化传输功率，以最大化节点寿命；以及混合声光中继（HAOR），这是一种双模式中继方案，将声学链路与用于高速近距离数据传输的光链路集成在一起，实现了极远距离可靠性。实验评估证明，与基线方法相比，AquaSense-AMC节省了28% %的能源，提高了35% %的数据包传输率，并使网络寿命延长了22% %。该模型实现了一种可持续的自适应通信系统，使IoUT在复杂的水下环境中可靠、节能地运行。

{"title":"AquaSense-AMC: An adaptive modulation-control model for energy-efficient communication in underwater IoT networks","authors":"S.R. Janani , Poornima. S , Aruna R , Chandru Vignesh C","doi":"10.1016/j.suscom.2025.101263","DOIUrl":"10.1016/j.suscom.2025.101263","url":null,"abstract":"<div><div>The Internet of Underwater Things (IoUT) is revolutionizing maritime research, environmental monitoring and intelligent aquatic applications with real-time sensing and communication. Energy-efficient communication is the biggest challenge because of the constraint of bandwidth, long delay of propagation and highly dynamic water channels. To combat these issues, AquaSense-AMC is introduced a new Adaptive Modulation-Control (AMC) Model for underwater IoT networks. The framework provides three novel components: Channel-Aware Modulation Switching (CAMS), dynamically switching between modulation depth and symbol rate based on channel fluctuations; Energy-Constrained Control Mechanism (ECCM), energy-optimizing transmission power with predictive energy management to maximize node lifetime; and Hybrid Acoustic-Optical Relay (HAOR), a two-mode relay scheme that integrates acoustic links for extreme distance reliability with optical links for high-rate near-distance data transfer. Experimental assessments prove that AquaSense-AMC saves energy by 28 %, enhances packet delivery ratio by 35 % and increases network lifetime by 22 % relative to baseline methods. The model implements a sustainable and adaptive communication system and making IoUT operation reliable and energy-efficient in intricate underwater environments.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101263"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884065","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

E2SRP: Energy efficient secure routing protocol for edge-assisted wireless sensor networks E2SRP：边缘辅助无线传感器网络的节能安全路由协议

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101287

T.N. Prabhu , C. Ranjeeth Kumar , B. Sabeena , Huda Fatima

Wireless Sensor Networks (WSNs) enable large-scale data collection through spatially distributed sensor nodes but face challenges due to limited energy, computational capacity, and security vulnerabilities. Traditional routing protocols optimized for wired networks are unsuitable for such constrained environments. This paper presents an Energy Efficient Secure Routing Protocol (E2SRP) for edge-assisted WSNs, addressing both energy consumption and security concerns. Trust values are computed using the Analytical Hierarchy Process (AHP) based on Direct, Indirect, and Witness Trust metrics to ensure reliable node selection. Optimal routing paths are identified through the Grey Wolf Optimization (GWO) algorithm, while BLAKE3 hashing performs secure data aggregation and deduplication at the edge. Additionally, Fuzzy Intelligence supports load balancing to enhance system stability. Simulation results using NS-3 demonstrate that the proposed model significantly improves Packet Delivery Ratio (PDR), reduces energy consumption and routing overhead, and strengthens overall security resilience compared with existing methods.

无线传感器网络（wsn）能够通过空间分布的传感器节点进行大规模数据采集，但由于能量、计算能力和安全漏洞的限制，它面临着挑战。传统的有线网络优化路由协议不适合这种约束环境。本文提出了一种用于边缘辅助wsn的节能安全路由协议（E2SRP），解决了能耗和安全问题。使用基于直接信任、间接信任和证人信任度量的层次分析法（AHP）计算信任值，以确保可靠的节点选择。采用灰狼优化算法（GWO）识别最优路由路径，采用BLAKE3哈希算法在边缘进行安全的数据聚合和重复数据删除。此外，模糊智能支持负载均衡，增强系统稳定性。基于NS-3的仿真结果表明，与现有方法相比，该模型显著提高了包投递率（PDR），降低了能耗和路由开销，增强了整体安全弹性。

{"title":"E2SRP: Energy efficient secure routing protocol for edge-assisted wireless sensor networks","authors":"T.N. Prabhu , C. Ranjeeth Kumar , B. Sabeena , Huda Fatima","doi":"10.1016/j.suscom.2025.101287","DOIUrl":"10.1016/j.suscom.2025.101287","url":null,"abstract":"<div><div>Wireless Sensor Networks (WSNs) enable large-scale data collection through spatially distributed sensor nodes but face challenges due to limited energy, computational capacity, and security vulnerabilities. Traditional routing protocols optimized for wired networks are unsuitable for such constrained environments. This paper presents an Energy Efficient Secure Routing Protocol (E2SRP) for edge-assisted WSNs, addressing both energy consumption and security concerns. Trust values are computed using the Analytical Hierarchy Process (AHP) based on Direct, Indirect, and Witness Trust metrics to ensure reliable node selection. Optimal routing paths are identified through the Grey Wolf Optimization (GWO) algorithm, while BLAKE3 hashing performs secure data aggregation and deduplication at the edge. Additionally, Fuzzy Intelligence supports load balancing to enhance system stability. Simulation results using NS-3 demonstrate that the proposed model significantly improves Packet Delivery Ratio (PDR), reduces energy consumption and routing overhead, and strengthens overall security resilience compared with existing methods.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101287"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145884064","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

AI-driven energy management for resilient operation of renewable-powered microgrids 人工智能驱动的可再生微电网弹性运行能源管理

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101288

Zhixiang Dai, Feng Wang, Shaoxiong Zhang, Li Xu

The microgrids function as distributed energy systems using renewable energy sources such as solar, wind, and energy storage. Microgrids have tremendous possibilities to improve an energy system in terms of resilience and sustainability. This paper essentially deals with describing an AI-based energy management system tailored for optimizing the operation of the renewable energy-powered microgrid. The EMS uses advanced ML techniques consisting of LSTM networks for forecasting, DQN for decision-making, and Random Forest depending upon solar energy forecasting. The models were tested on several benchmark datasets, namely, renewable, solar, wind, and a global energy dataset. According to the results, the model with LSTM gave the best forecasts, particularly for the Wind Energy Dataset (R ² = 0.9955). On the other hand, the highest performance for the Solar Energy Dataset was obtained by a random forest, which gave it a lesser predictive power radius than other techniques (R² = 0.827). The other energy datasets also witness optimal working of LSTM time-series approaches, giving much smaller errors for training and validation phases. Moreover, AI-assisted EMS, given improved energy resource optimization, forecasting, and operational resilience, will play a vital role in managing renewable microgrids under dynamically ever-changing and uncertain environments. This study also endorses the application of AI in furthering and passing good genes of efficiency, sustainability, and fault tolerance into renewable-powered microgrids.

微电网作为分布式能源系统，使用太阳能、风能和储能等可再生能源。微电网在改善能源系统的弹性和可持续性方面具有巨大的可能性。本文主要描述了一种基于人工智能的能源管理系统，该系统是为优化可再生能源微电网的运行而量身定制的。EMS使用先进的机器学习技术，包括用于预测的LSTM网络，用于决策的DQN和依赖于太阳能预测的随机森林。这些模型在几个基准数据集上进行了测试，即可再生能源、太阳能、风能和全球能源数据集。根据结果，LSTM模型给出了最好的预测，特别是对风能数据集（r2 = 0.9955）。另一方面，太阳能数据集的最高性能是由随机森林获得的，这使得它的预测能力半径比其他技术要小（R²= 0.827）。其他能源数据集也见证了LSTM时间序列方法的最佳工作，为训练和验证阶段提供了更小的误差。此外，人工智能辅助的能源管理系统，由于改进了能源资源优化、预测和运行弹性，将在动态变化和不确定环境下管理可再生微电网方面发挥至关重要的作用。这项研究还支持了人工智能在可再生能源微电网中进一步推广和传递效率、可持续性和容错的良好基因的应用。

{"title":"AI-driven energy management for resilient operation of renewable-powered microgrids","authors":"Zhixiang Dai, Feng Wang, Shaoxiong Zhang, Li Xu","doi":"10.1016/j.suscom.2025.101288","DOIUrl":"10.1016/j.suscom.2025.101288","url":null,"abstract":"<div><div>The microgrids function as distributed energy systems using renewable energy sources such as solar, wind, and energy storage. Microgrids have tremendous possibilities to improve an energy system in terms of resilience and sustainability. This paper essentially deals with describing an AI-based energy management system tailored for optimizing the operation of the renewable energy-powered microgrid. The EMS uses advanced ML techniques consisting of LSTM networks for forecasting, DQN for decision-making, and Random Forest depending upon solar energy forecasting. The models were tested on several benchmark datasets, namely, renewable, solar, wind, and a global energy dataset. According to the results, the model with LSTM gave the best forecasts, particularly for the Wind Energy Dataset (R <sup>2</sup> = 0.9955). On the other hand, the highest performance for the Solar Energy Dataset was obtained by a random forest, which gave it a lesser predictive power radius than other techniques (R² = 0.827). The other energy datasets also witness optimal working of LSTM time-series approaches, giving much smaller errors for training and validation phases. Moreover, AI-assisted EMS, given improved energy resource optimization, forecasting, and operational resilience, will play a vital role in managing renewable microgrids under dynamically ever-changing and uncertain environments. This study also endorses the application of AI in furthering and passing good genes of efficiency, sustainability, and fault tolerance into renewable-powered microgrids.</div></div>","PeriodicalId":48686,"journal":{"name":"Sustainable Computing-Informatics & Systems","volume":"49 ","pages":"Article 101288"},"PeriodicalIF":5.7,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Retraction notice to “Energy-efficient blockchain-integrated IoT and AI framework for sustainable urban microclimate management” [Sustain. Comput.: Inf. Syst. 47 (2025) 101137] 关于“节能区块链集成物联网和人工智能框架用于可持续城市微气候管理”的撤回通知[…]第一版。[参考文献47 (2025)101137]

IF 5.7 3区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Sustainable Computing-Informatics & Systems

Pub Date : 2026-01-01 DOI: 10.1016/j.suscom.2025.101227

N. Krishnaraj , Hadeel Alsolai , Fahd N. Al-Wesabi , Yahia Said , Ali Alqazzaz , S. Gayathri Priya , S. Shanmathi , B. Narmada

引用次数: 0