IEEE Transactions on Sustainable Computing最新文献_第2页

Motion-Aware Medium Access Control With Priority-Driven Resource Allocation for Wireless Body Area Networks 具有优先级驱动的无线体域网络的运动感知介质访问控制

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-09-01 Epub Date: 2026-07-20 DOI: 10.1109/TSUSC.2026.3715180

Yurui Zhang;Xuxun Liu;Huan Zhou;Victor C. M. Leung

{"title":"Motion-Aware Medium Access Control With Priority-Driven Resource Allocation for Wireless Body Area Networks","authors":"Yurui Zhang;Xuxun Liu;Huan Zhou;Victor C. M. Leung","doi":"10.1109/TSUSC.2026.3715180","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3715180","url":null,"abstract":"Medium Access Control (MAC) is critical for ensuring reliable and timely transmission of physiological data in Wireless Body Area Networks (WBANs). However, existing MAC designs fail to address tripartite challenges: priority-agnostic contention in dense networks, rigid resource allocation under dynamic channel conditions, and disjointed retransmission and resource scheduling mechanisms. This paper proposes a motion-aware priority-driven MAC (MAPD-MAC) protocol to overcome these limitations. The protocol synergizes contention-based and contention-free channel access mechanisms through three innovations: a priority-isolated channel access framework that eliminates cross-priority collisions and mitigates energy waste through dynamic contention windows, a body-motion-aware Markov model enabling adaptive packet sizing and slot allocation aligned with predicted channel stability, and a context-adaptive retransmission scheduler that prioritizes critical data while deferring non-urgent transmissions during fading episodes. Three distinctive features further distinguish this work: urgency-driven channel access mode switching, cross-layer interference management, and clinically integrated resource orchestration. Extensive simulations demonstrate significant improvements over state-of-the-art protocols, achieving a higher packet success rate, lower energy consumption, and reduction in average Age of Information (AoI) under dynamic and dense WBANs. These advancements collectively bridge the gap between stringent clinical demands and network resource constraints, propelling WBANs toward responsive and sustainable healthcare applications.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 5","pages":"570-581"},"PeriodicalIF":5.3,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149131235","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

Modeling Scenarios for Carbon-Aware Geographic Load Shifting of Compute Workloads 计算工作负载的碳感知地理负载转移建模方案

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-06-10 DOI: 10.1109/TSUSC.2026.3702249

Wim Vanderbauwhede

引用次数: 0

Chunk-Based Distributed Tensor-Train Decomposition Methods for Cyber-Physical-Social Intelligence 基于块的网络-物理-社会智能分布式张量序列分解方法

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2025-07-10 DOI: 10.1109/TSUSC.2025.3585583

Xiaokang Wang;Kuining Feng;Laurence T. Yang;Nenggan Zheng;M. Jamal Deen

{"title":"Chunk-Based Distributed Tensor-Train Decomposition Methods for Cyber-Physical-Social Intelligence","authors":"Xiaokang Wang;Kuining Feng;Laurence T. Yang;Nenggan Zheng;M. Jamal Deen","doi":"10.1109/TSUSC.2025.3585583","DOIUrl":"https://doi.org/10.1109/TSUSC.2025.3585583","url":null,"abstract":"<inline-formula><tex-math>$mathbf {C}$</tex-math></inline-formula>yber-<inline-formula><tex-math>$mathbf {P}$</tex-math></inline-formula>hysical-<inline-formula><tex-math>$mathbf {S}$</tex-math></inline-formula>ocial <inline-formula><tex-math>$mathbf {S}$</tex-math></inline-formula>ystems (CPSS), emerging as a new paradigm of computing applications, combine <inline-formula><tex-math>$mathbf {C}$</tex-math></inline-formula>yber-<inline-formula><tex-math>$mathbf {P}$</tex-math></inline-formula>hysical <inline-formula><tex-math>$mathbf {S}$</tex-math></inline-formula>ystems (CPS) with the social characteristics for the Cyber-Physical-Social Intelligence to provide the services at the right time, the right place and by the right means. For this purpose, CPSS Big Data should be processed and analyzed efficiently to explore the hidden information of users. As an effective processing tool of CPSS Big Data from the perspective of multi-attributes, tensor with its decomposition methods has attracted much attention. At the same time, <inline-formula><tex-math>$mathbf {t}$</tex-math></inline-formula>ensor-<inline-formula><tex-math>$mathbf {t}$</tex-math></inline-formula>rain (TT), a potential tensor decomposition method, has proven to be an efficient CPSS data analysis method. In this paper, as the extension of our previous work, chunk-based distributed tensor-train decomposition methods with the tensor divided in the form of blocks are proposed to speed up data processing and improve the adaptability and extensibility. Finally, experiments carried out on the random data and medical image data are used to measure the performance of the proposed methods including the error and computational efficiency.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"350-361"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699574","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

Enhancing Multi-Step Solar Irradiance Forecasting Through TVF-EMD-AWNN: A Time-Varying Hybrid Framework for PV Power Plant 利用TVF-EMD-AWNN增强多步太阳辐照度预测：光伏电站时变混合框架

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-03-08 DOI: 10.1109/TSUSC.2026.3688713

Sakshi Shukla;Sumanta Pasari;Kshitij Gupta

{"title":"Enhancing Multi-Step Solar Irradiance Forecasting Through TVF-EMD-AWNN: A Time-Varying Hybrid Framework for PV Power Plant","authors":"Sakshi Shukla;Sumanta Pasari;Kshitij Gupta","doi":"10.1109/TSUSC.2026.3688713","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3688713","url":null,"abstract":"Accurate solar irradiance prediction is inevitable for estimating photovoltaic energy generation and meeting utility demands. This paper presents a novel TVF-EMD-AWNN hybrid framework that integrates time-varying filter empirical mode decomposition (TVF-EMD) with an adaptive wavelet neural network (AWNN) for enhanced prediction accuracy at an hourly timescale. The proposed framework encompasses a sophisticated three-fold practically relevant integration architecture. First, the adaptive filtering mechanism of TVF enables dynamic frequency-time optimization while maintaining non-stationary signal characteristics through its time-varying filtering process. Then, the EMD decomposes these enhanced signals into intrinsic mode functions, exhibiting unique temporal-spectral signatures and facilitating comprehensive multiresolution analysis. Finally, the adaptive wavelet parameters in AWNN synergistically represent these decomposed components through dynamic scale-translation optimization in the wavelet space. Additionally, the proposed TVF-EMD-AWNN model is also validated through the expanding window cross-validation approach. The theory is illustrated on a long hourly solar irradiance data (2015–2021) at four diverse climatic regions of India. Results demonstrate noteworthy accuracy improvement in terms of relative RMSE reduction in comparison to established standalone and hybrid models, such as MLP (83%), LSTM (85%), Bi-LSTM (82%), EDBi-LSTM (80%), AWNN (78%), CEEMDAN-LSTM (83%), CEEMDAN-EDBi-LSTM (80%), TVF-EMD-LSTM (65%), and TVF-EMD-BiLSTM (54%). Besides, the performance of TVF-EMD-AWNN remains consistent even at multi-step hourly predictions. Therefore, the proposed hybrid framework with additional benchmarking, ablation studies, and time series validation offers robustness towards practical power system operation.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"376-387"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699477","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

Sustainability via Large Language Model Right-Sizing 通过大语言模型调整大小实现可持续性

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-06-12 DOI: 10.1109/TSUSC.2026.3697261

Jennifer Haase;Finn Klessascheck;Jan Mendling;Sebastian Pokutta

{"title":"Sustainability via Large Language Model Right-Sizing","authors":"Jennifer Haase;Finn Klessascheck;Jan Mendling;Sebastian Pokutta","doi":"10.1109/TSUSC.2026.3697261","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3697261","url":null,"abstract":"Large language models (LLMs) are increasingly embedded in organizational workflows, raising concerns about energy use, costs, and data sovereignty. While benchmarks often celebrate cutting-edge models, real-world deployment requires a broader question: when is a smaller, locally deployable model “good enough”? This study evaluates eleven proprietary and open-weight LLMs on ten everyday occupational tasks, including summarizing, scheduling, and drafting emails and proposals. Using a dual-LLM evaluation framework, we automated task execution and standardized assessment across ten criteria of quality, accuracy, and ethical responsibility. Results show GPT-4o delivers top performance but with significantly higher costs and environmental impact. Smaller models such as Gemma-3 and Phi-4 performed strongly on most tasks, supporting their viability where cost-efficiency, local use, or privacy matter. A cluster analysis identified three groups—High, Balanced, and Baseline Performance Models—highlighting trade-offs between quality, control, and sustainability. Task type shaped effectiveness: conceptual tasks were most challenging, while aggregation and transformation tasks yielded better outcomes. We argue for shifting from performance-maximizing benchmarks to task- and context-aware sufficiency assessments that reflect organizational priorities. Our approach offers a replicable method to evaluate AI models from a sustainability perspective and provides practical recommendations for responsible LLM deployment.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"426-438"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11561102","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699573","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Data-Driven Convex Relaxation for Sustainable Optimal Power Flow With Feasibility Guarantees 具有可行性保证的可持续最优潮流的数据驱动凸松弛

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-04-28 DOI: 10.1109/TSUSC.2026.3688521

Junfei Wang;Pirathayini Srikantha

{"title":"Data-Driven Convex Relaxation for Sustainable Optimal Power Flow With Feasibility Guarantees","authors":"Junfei Wang;Pirathayini Srikantha","doi":"10.1109/TSUSC.2026.3688521","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3688521","url":null,"abstract":"The sustainable optimal power flow (SOPF) problem is formulated by grid operators to jointly optimize cost efficiency and carbon emissions, while adhering to the operational limits and constraints of the underlying electrical infrastructure. With the proliferation of intermittent generation sources (e.g. wind and solar power) and diverse demands (e.g. electric vehicles), SOPF must be solved at high granularity to account for the variabilities introduced by these. However, due to the nonconvex nature of SOPF, solving it directly is NP-hard. In this paper, we introduce a data-driven convex relaxation framework that approximates the decision boundary separating feasible and infeasible points of the SOPF problem. This convex boundary replaces all nonlinear constraints in the formulation. This transformed problem is then solved iteratively via a gradient-based approach for which the solution is globally optimal. Then, in order to guarantee feasibility for the original problem, a simplified physics-based refinement is introduced and its global convergence properties are proven using theoretical constructs from fixed point iteration. Practical and comparative case studies on standard benchmarks, including IEEE 14-bus, IEEE 118-bus, and PEGASE 1354-bus systems, demonstrate the effectiveness of the proposed approach in terms of optimality, feasibility, and sustainability.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"362-375"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699429","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

Robust Multi-Level Forecast-Based Anomaly Detection for Smart Grid Overload Attacks 基于多级预测的智能电网过载异常检测

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-03-25 DOI: 10.1109/TSUSC.2026.3696332

Thamidu Naveen;Malka N. Halgamuge;Sisil Kumarawadu;Shahriar Kaisar;V. Logeeshan

{"title":"Robust Multi-Level Forecast-Based Anomaly Detection for Smart Grid Overload Attacks","authors":"Thamidu Naveen;Malka N. Halgamuge;Sisil Kumarawadu;Shahriar Kaisar;V. Logeeshan","doi":"10.1109/TSUSC.2026.3696332","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3696332","url":null,"abstract":"Smart grids rely on advanced metering infrastructure (AMI) for real-time usage data, yet this connectivity introduces vulnerabilities. One critical threat is grid overloading cyberattacks, where an adversary manipulates demand to exceed safe limits and trigger blackouts. Such attacks can harness compromised smart meters to drastically raise the load and destabilize the network. Despite their severity, AMI data manipulation for overloading attacks has received limited attention. We propose a novel anomaly detection framework that combines household and neighborhood-level load predictions. Using deep learning models, our system predicts the electricity usage of each customer, as well as the aggregate community load, allowing detection of sudden deviations that signal an attack. To quantify anomalies, we introduce two metrics: an Abnormality Index (capturing the magnitude of peak deviations) and a Regularity Index (measuring energy consumption pattern consistency). These indices are fed into an ensemble model to detect threats. Crucially, our ensemble model combines multiple classifiers to mitigate adversarial manipulation, making it robust to data poisoning and evasion attempts. Extensive evaluation with a real dataset demonstrates our proposed model achieves a detection rate of 93% and a false alarm rate of 9% in the most challenging baseline conditions, and maintains 81% detection even under the hardest adversarial setting using glass-boxAutoAttack scenarios, outperforming existing contemporary methods. This integrated solution advances smart grid cybersecurity by combining predictive load modeling, novel anomaly metrics, and robust classification.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"414-425"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699546","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

ITD-Aware CPU Power Modeling to Identify Peak-Efficiency Thermal Regions for Sustainable Data Center Operation 可识别过渡段的CPU功率建模，以确定可持续数据中心运行的峰值效率热区

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-03-20 DOI: 10.1109/TSUSC.2026.3695348

Jason Crop;Hayden Moore;Sudeep Pasricha

{"title":"ITD-Aware CPU Power Modeling to Identify Peak-Efficiency Thermal Regions for Sustainable Data Center Operation","authors":"Jason Crop;Hayden Moore;Sudeep Pasricha","doi":"10.1109/TSUSC.2026.3695348","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3695348","url":null,"abstract":"Data center power consumption continues to grow rapidly with the increasing demand for cloud and enterprise computing. CPU active power and cooling account for a major share of this energy use, yet manufacturers provide no visibility into each CPU’s unique power–thermal behavior. Process variations and inverse temperature dependence (ITD) cause individual CPUs to exhibit distinct electrical characteristics — including voltage, dynamic capacitance, and leakage — which together determine their optimal operating temperature for maximum performance-per-watt. This paper introduces a new ITD-aware CPU power modeling framework that integrates empirical voltage, leakage, and capacitance measurements to predict total power across temperature and frequency with high accuracy. The model is validated on 22 Intel Xeon server CPUs, achieving less than 3% error across operating conditions. Using the model, we identify each CPU’s optimal temperature that minimizes power while sustaining performance. Analysis of real cloud servers reveals that around 50% of modern CPUs operate roughly 10 °C below their optimal point, wasting approximately 15% in excess cooling energy. Classifying CPUs by their optimal temperature enables data centers to group and cool them more efficiently, reducing energy use and improving sustainability without performance degradation.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"388-401"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699478","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

UlpFIFO: A 32nm 8-Bit 5.2-$mathrm{mu }$μW Asynchronous FIFO With Hybrid Power Management for Edge Computing UlpFIFO：一种32纳米8位5.2-$ mathm {mu}$μW异步FIFO与混合电源管理的边缘计算

IF 5.3 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-07-01 Epub Date: 2026-06-15 DOI: 10.1109/TSUSC.2026.3703817

Subhadeep Nag;Suman Kalyan Porel;Aniruddha Chandra;Hemanta Kumar Mondal

{"title":"UlpFIFO: A 32nm 8-Bit 5.2-$mathrm{mu }$μW Asynchronous FIFO With Hybrid Power Management for Edge Computing","authors":"Subhadeep Nag;Suman Kalyan Porel;Aniruddha Chandra;Hemanta Kumar Mondal","doi":"10.1109/TSUSC.2026.3703817","DOIUrl":"https://doi.org/10.1109/TSUSC.2026.3703817","url":null,"abstract":"In response to the escalating demand for low-power and high-performance digital architectures, the design of energy-efficient data buffering mechanisms has emerged as a critical area of focus, particularly within edge computing, artificial intelligence accelerators, and high-speed communication systems. Asynchronous memory buffers facilitate seamless data transfer across heterogeneous clock domains. However, conventional implementations are often plagued by considerable power inefficiencies. This work introduces a novel architectural enhancement of an 8-bit asynchronous Ultra Low Power FIFO (UlpFIFO), employing a hybrid power management (HPM) strategy that synergistically integrates clock gating and power gating techniques. This dual-pronged approach effectively mitigates both dynamic and leakage power dissipation. The proposed design is rigorously evaluated through implementation on both FPGA and 32 nm CMOS ASIC platforms. Empirical results reveal a 25% reduction in dynamic power consumption, accompanied by a 12% decline in leakage power, at the cost of a modest 2.5% increase in silicon area.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 4","pages":"439-452"},"PeriodicalIF":5.3,"publicationDate":"2026-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148699488","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

Importance Aware Undervolting for Robust Neural Network Training 感知欠电压对鲁棒神经网络训练的重要性

IF 3.9 3区计算机科学

IEEE Transactions on Sustainable Computing Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI: 10.1109/TSUSC.2025.3650602

Chen Zhang;Lening Wang;Xin Fu

{"title":"Importance Aware Undervolting for Robust Neural Network Training","authors":"Chen Zhang;Lening Wang;Xin Fu","doi":"10.1109/TSUSC.2025.3650602","DOIUrl":"https://doi.org/10.1109/TSUSC.2025.3650602","url":null,"abstract":"Convolutional Neural Network (CNN) is a powerful tool that has been extensively applied to many different applications. However, recent developments at CNN have revealed its vulnerability against adversarial example attacks. By introducing visually undetectable noise to the input image, an adversarial example attack can cause the CNN classifier to make false predictions. Multiple approaches have been proposed to defend against adversarial samples, one of which focuses on injecting noise into CNN during training. However, the existing method cannot generate noise efficiently and introduces extra time and energy overhead. In this paper, we propose an Importance-Aware undervolting training framework to improve CNN robustness. The undervolting technique is employed during training for noise generation at negligible overhead. Meanwhile, we observe that the neuron importance and bit importance in hardware can be leveraged during undervolting CNN training for controllable and flexible noise injection, which improves robustness. We design a position-aware bit mapping method in the memory unit by allocating data bits based on significance. And an importance-aware processing element (PE) mapping is also proposed at the computation unit for noise restriction. Our approach regularizes the noise injected into CNN and serves as an efficient method to defend against adversarial example attacks with significant energy savings. The proposed framework is evaluated through both FPGA implementation and software simulation. The experiment results show that our importance-aware undervolting CNN training achieves 47.8% adversarial accuracy at PGD-10 attack and 47.0% training energy savings.","PeriodicalId":13268,"journal":{"name":"IEEE Transactions on Sustainable Computing","volume":"11 2","pages":"147-157"},"PeriodicalIF":3.9,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147606299","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