Collaborative Research: NGSDI: CarbonFirst: A Sustainable and Reliable Carbon-Centric Cloud-Edge Software Infrastructure

合作研究：NGSDI：CarbonFirst：可持续且可靠的以碳为中心的云边缘软件基础设施

• 批准号: 2105648
• 负责人: Adam Wierman
• 金额: $ 36.11万
• 依托单位: California Institute of Technology
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-05-01 至 2025-04-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2105648&HistoricalAwards=false
• 关键词: Collaborative; Research; NGSDI; CarbonFirst; Sustainable

Cloud computing platforms continue to grow exponentially, and are becoming the foundation of our information-based economy. While the cloud’s energy demand grew more slowly than expected over the past decade due to aggressive energy-efficiency optimizations, there are few remaining optimization opportunities using traditional methods. As a result, the cloud’s continued exponential growth will translate into exponentially rising energy demand, which will position it as one of the primary contributors to global carbon emissions. To address the problem, this project elevates carbon to a first-class metric in designing a sustainable and reliable cloud-edge software infrastructure that can enable continued exponential growth.The project's foundation is a software-defined energy virtualization layer that provides applications visibility into, and control of, their own energy and carbon usage. The project will leverage this foundation to develop higher-level systems abstractions for supporting carbon-efficient applications at different geographical scales including: a cluster balloon technique, which automatically adjusts applications’ energy usage to match a volatile clean energy supply at local edge sites; edge hopping mechanisms, which exploit lower regional energy volatility to balance energy across edge sites; and carbon capping policies, which track applications’ global grid carbon emissions and restrict grid energy after reaching the cap.The project has the potential for significant societal impact by enabling commercial cloud platforms to sustainably continue their exponential growth. The project will conduct outreach by incorporating topics from the proposal into summer programs for local middle and high school students at the partner institutions. The project will also impact the curriculum at these institutions by adopting elements of edge, cloud, and sustainable computing into graduate and advanced undergraduate courses. Finally, the project will recruit a diverse group of students by leveraging institutional diversity efforts and will involve undergraduate students through Research Experience for Undergraduate (REU) projects.The project will make its software artifacts, datasets, and research results available to the research community on the project website at http://www.carbonfirst.org and via the UMass Trace Repository at http://traces.cs.umass.edu. Artifacts derived from this project will be maintained on the project website and the trace repository for a minimum of five years after the project's conclusion.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

云计算平台继续呈指数级增长，并正在成为我们基于信息的经济的基础，尽管由于积极的能源效率优化，云的能源需求在过去十年中增长速度比预期慢，但使用传统技术的优化机会所剩无几。因此，云的持续指数增长将转化为能源需求的指数增长，这将使其成为全球碳排放的主要贡献者之一。为了解决这个问题，该项目将碳提升到了一流的指标。设计一个可持续且可靠的可以实现持续指数增长的云边缘软件基础设施。该项目的基础是软件定义的能源虚拟化层，为应用程序提供对其自身能源和碳使用的可见性和控制。该项目将利用这一基础来开发更高的。支持不同地理范围内的碳高效应用的水平系统抽象，包括：集群气球技术，自动调整应用程序的能源使用，以匹配本地边缘站点不稳定的清洁能源供应；边缘跳跃机制，利用较低的区域能源波动性平衡边缘站点的能源；和碳上限政策，跟踪应用程序的全球电网碳排放并在达到上限后限制电网能源。该项目通过使商业云平台可持续地持续指数增长，具有重大社会影响的潜力。该项目将通过纳入来自主题的主题进行推广。该项目还将通过在研究生和高级本科生课程中采用边缘、云和可持续计算元素来影响这些机构的课程。通过利用机构多元化的努力来吸引多元化的学生群体，并将通过以下方式让本科生参与进来本科生研究经验 (REU) 项目。该项目将通过项目网站 http://www.carbonfirst.org 以及通过 UMass Trace 存储库 (http://www.carbonfirst.org) 向研究界提供其软件工件、数据集和研究结果： //traces.cs.umass.edu。该项目得出的工件将在项目结束后在项目网站和跟踪存储库上保留至少五年。该奖项反映了 NSF 的法定使命，并被认为值得支持。通过使用基金会的智力优点和更广泛的影响审查标准进行评估。

项目成果

Scalable Reinforcement Learning for Multiagent Networked Systems

多智能体网络系统的可扩展强化学习

• DOI: 10.1287/opre.2021.2226
• 发表时间: 2022-02
• 期刊: Operations Research
• 影响因子: 2.7
• 作者: Qu, Guannan;Wierman, Adam;Li, Na
• 通讯作者: Li, Na

Smoothed Online Optimization with Unreliable Predictions

通过不可靠的预测进行平滑的在线优化

• DOI: 10.1145/3579442
• 发表时间: 2022-02-07
• 期刊: Proceedings of the ACM on Measurement and Analysis of Computing Systems
• 作者: Daan Rutten;Nicolas H. Christianson;Debankur Mukherjee;A. Wierman
• 通讯作者: A. Wierman

Pareto-Optimal Learning-Augmented Algorithms for Online Conversion Problems

在线转换问题的帕累托最优学习增强算法

• 发表时间: 2021-01
• 期刊: Advances in Neural Information Processing Systems 34 (NeurIPS 2021
• 作者: Bo Sun; Russell Lee
• 通讯作者: Russell Lee

Minimization Fractional Prophet Inequalities for Sequential Procurement

顺序采购的分数先知不等式最小化

• DOI: 10.1287/moor.2021.173
• 发表时间: 2023-06
• 期刊: Mathematics of Operations Research
• 影响因子: 1.7
• 作者: Qin, Junjie;Vardi, Shai;Wierman, Adam
• 通讯作者: Wierman, Adam

Online Optimization with Feedback Delay and Nonlinear Switching Cost

具有反馈延迟和非线性切换成本的在线优化

• DOI: 10.1145/3508037
• 发表时间: 2022-02
• 期刊: Proceedings of the ACM on Measurement and Analysis of Computing Systems
• 作者: Pan, Weici;Shi, Guanya;Lin, Yiheng;Wierman, Adam
• 通讯作者: Wierman, Adam