CAREER: An Automated Compiler-Runtime Framework for Democratizing Secure Collaborative Computation

最近在高级分析和机器学习方面的革命引起了组织对高价值数据日益增长的需求。然而，在许多领域，获得足够数量的此类数据是具有挑战性的，因为大多数高价值数据是敏感的，并且由多个组织拥有并分散在多个组织中。由于隐私问题、监管政策和/或商业竞争，组织通常无法共享他们拥有的数据。一个很有前途的解决方案是安全多方计算（MPC），这是一种加密技术，允许组织在其联合数据集上运行复杂的计算，而不会向其他方泄露敏感输入。不幸的是，开发高效的协议仍然是一个劳动密集型的过程，需要加密专业知识，对于大多数用户和开发人员来说是遥不可及的。该项目更广泛的意义和重要性在于，通过设计和构建端到端的集成编译器运行时框架，可以自动生成和执行优化的、特定于工作负载的MPC协议，加速MPC的普及。该项目的新颖之处在于通过开发和结合来自编译器、密码学和分布式系统的新见解来构建这样一个框架。研究小组计划从三个方面来解决这个研究问题。第一个重点是通过支持可扩展到各种MPC原语的基本编译基础设施来开发平台的基础。第二个推力为编译器设计了新的技术，为用户提供的高级程序生成特定于工作负载的MPC协议。第三个推力创建了一个新的分布式运行时，用于有效地执行和调度使用推力2中的技术生成的MPC协议。该团队正计划将该框架开源给其他研究人员和教育工作者，并正在努力创建新的研究生课程和推广项目，使本科生更容易获得研究。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 recent revolution in advanced analytics and machine learning gave rise to a growing demand among organizations for high-value data. However, obtaining sufficient volumes of such data is challenging in many areas because most high-value data is sensitive, and is owned by and split across multiple organizations. Organizations often cannot share the data they own due to privacy concerns, regulatory policies, and/or business competition. A promising solution is secure multiparty computation (MPC), a cryptographic technique that allows organizations to run complex computations on their joint dataset without revealing sensitive inputs to other parties. Unfortunately, developing efficient protocols remains a labor-intensive process that requires cryptographic expertise and is out of reach for most users and developers. The project's broader significance and importance are to accelerate and democratize the adoption of MPC by designing and building an end-to-end, integrated compiler-runtime framework that automatically generates and executes optimized, workload-specific MPC protocols. The project’s novelties are to build such a framework by developing and combining new insights from compilers, cryptography, and distributed systems. The team of researchers plan to address the research question in three broad thrusts. The first thrust develops the foundation of the platform by supporting a basic compilation infrastructure that is extensible to a wide variety of MPC primitives. The second thrust designs novel techniques for the compiler to generate workload-specific MPC protocols for user-provided high-level programs. The third thrust creates a new distributed runtime for efficiently executing and scheduling MPC protocols that are generated using techniques from Thrust 2. The team is planning to open source the framework for other researchers and educators, and is working to create new graduate courses and outreach programs to make research more accessible to undergraduate students.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.