CAREER: A Robust and Data-driven Design for Carbon-intelligent Distributed Systems

互联网是全天候的服务，它也可以是全天候无碳的。这是一个雄心勃勃的目标，作为计算机领域对气候变化的回应，互联网先驱行业最近一直在倡导这个目标。实现这一目标需要对地理分布的数字基础设施（如数据中心、提供视频流等服务或托管计算密集型应用程序）的操作进行脱碳。尽管在数字基础设施的可持续性方面做了十年的努力，但目前的设计仍然不足以实现24/7无碳的目标。目前的设计重点是最小化能源成本和/或消耗或最大化可再生能源的参与。后者是最接近脱碳目标的努力；然而，当风不吹或太阳不照耀时，这种设计没有碳智能，即它忽略了电网的能量，而没有考虑到电网的碳强度随着时间的推移而变化，因为存在其他无碳来源，如水电。该方案明确将碳智能作为数字基础设施的一流设计原则。它将开发和评估新的碳智能方法，这些方法对不确定性具有鲁棒性，并为数据驱动的适应提供设计空间，以提高实际性能。所提出的理论将为设计健壮和数据驱动的系统奠定基础，并将更广泛地应用于计算机和网络系统。碳智能方法为推进计算机科学与可持续发展交叉领域的研究开辟了新的研究方向。更广泛地说，这个提议是计算机领域对气候变化的回应，气候变化是人类社会所面临的最大问题之一。最后，它为教育、社会技术和推广活动创造了机会。计划中的社会技术活动扩大了该提案的研究成果，以衡量和改善能源系统的不平等。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The internet is a 24/7 service, it could be 24/7 carbon-free too. This is an ambitious goal that has been recently advocated by internet pioneer industries as a response of the computing domain to climate change. Achieving this goal requires the decarbonization of operations of geographically distributed digital infrastructure, e.g., data centers, providing services such as video streaming, or hosting compute-intensive applications. Despite a decade of effort on the sustainability of digital infrastructure, the current design is still inadequate to achieve the 24/7 carbon-free goal. The current design focuses to minimize the energy cost and/or consumption or maximize renewable participation. The latter is the closest effort towards the decarbonization goal; however, when the wind is not blowing or the sun is not shining, this design has no carbon intelligence, i.e., it obliviously draws energy from the grid without taking into account that the carbon intensity of grid changes over time due to existence of other carbon-free sources such as hydropower. This proposal explicitly considers carbon intelligence as a first-class design principle of digital infrastructure. It will develop and evaluate novel carbon-intelligent methodologies that are robust against uncertainty and provide a design space for data-driven adaptation for improved practical performance. The proposed theories will create a foundation for designing robust and data-driven systems and will be applicable more broadly to the computer and network systems. The carbon-intelligence approaches open new research directions for advancing the research at the intersection of computer science and sustainability. More broadly, this proposal is a response of the computing domain to climate change, one of the largest problems’ society has ever faced. Last, it creates opportunities for educational, sociotechnical, and outreach activities. The planned sociotechnical activities broaden the research outcome of this proposal for measuring and improving the inequity of the energy system.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.