Collaborative Research: CSR: Medium: Architecting GPUs for Practical Homomorphic Encryption-based Computing

协作研究：CSR：中：为实用的同态加密计算构建 GPU

• 批准号: 2312276
• 负责人: Ajay Joshi
• 金额: $ 60万
• 依托单位: Trustees of Boston University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-08-01 至 2026-07-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2312276&HistoricalAwards=false
• 关键词: Collaborative; Research; CSR; Medium; Architecting

Cloud computing has become a popular path for efficiently sharing compute resources. However, the cloud can be an unsafe computing environment. To prevent data exposure, one can use Fully Homomorphic Encryption (FHE)-based computing in the cloud. FHE provides strong data privacy guarantees because it enables operations on encrypted data. Unfortunately, processing encrypted data using FHE takes multiple orders of magnitude longer than processing unencrypted data due to its prohibitively high compute and memory requirements. This project will explore the use of graphics processing units (GPUs) to accelerate FHE-based computing. We consider three different FHE schemes: Brakerski-Gentry-Vaikuntanathan (BGV), Brakerski/Fan-Vercauteren (B/FV), and Cheon-Kim-Kim-Song (CKKS), thus supporting operations on both integers and floating-point numbers.This project will advance the state-of-the-art in GPU compute and memory architectures to enable practical FHE-based computing in the cloud. We will also deliver new FHE benchmarks for GPUs and simulation tools. The outcomes of the proposed research will have a direct impact on the design of next-generation privacy-preserving computing systems. We will work with a network of companies to evaluate our work in a practical setting and disseminate it. We will also open-source software and tools resulting from our work to benefit the broader research community. We will actively participate in the Broadening Participation in Computing plans at both Northeastern University and Boston University, while developing a number of new programs to engage a diverse audience.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.

云计算已经成为有效共享计算资源的流行途径。然而，云计算可能是一个不安全的计算环境。为了防止数据泄露，可以在云中使用基于全同态加密（FHE）的计算。FHE提供强大的数据隐私保证，因为它支持对加密数据的操作。不幸的是，使用FHE处理加密数据比处理未加密数据要长几个数量级，这是由于其极高的计算和存储器要求。该项目将探索使用图形处理单元（GPU）来加速基于FHE的计算。我们考虑了三种不同的FHE方案：Brakerski-Gentry-Vaikuntanathan（BGV），Brakerski/Fan-Vercauteren（B/FV）和Cheon-Kim-Kim-Song（CKKS），从而支持整数和浮点数的操作。该项目将推进GPU计算和内存架构的最新技术，以实现基于FHE的实用云计算。我们还将为GPU和模拟工具提供新的FHE基准测试。所提出的研究成果将对下一代隐私保护计算系统的设计产生直接影响。我们将与一个公司网络合作，在实际环境中评估我们的工作并传播它。我们还将开放源代码软件和工具，使更广泛的研究社区受益。 我们将积极参与东北大学和波士顿大学的“扩大参与计算"计划，同时开发一些新项目以吸引不同的受众。该奖项反映了NSF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。