SaTC: CORE: Medium: Collaborative: New Approaches for Large Scale Secure Computation

安全多方计算（MPC）允许多个互不信任的各方在不向彼此公开数据的情况下对其数据执行联合计算。由于具有良好的安全性保证，MPC加密协议在过去十年中获得了前所未有的曝光。随着Tor和区块链等大型分布式系统和计算的激增，迫切需要大规模的MPC来确保这些应用程序的安全性和隐私性。然而，大规模应用对MPC协议的设计提出了一些独特的挑战。有成千上万的参与方，通常拥有不相等的资源，并且在协议执行期间，一些参与方可能会失败。本项目建议设计MPC协议来应对这些挑战。本课题的新颖之处在于设计了满足大规模分布式应用需求的新型MPC协议。具体而言，研究人员提出了一种新的基于委员会的大规模安全计算MPC框架。这些协议将计算分布到数千个并行委员会中，以最大限度地利用可用资源以提高吞吐量。所设计的协议优化选择委员会和委员会子协议，优化资源利用率和安全性。此外，它们还处理一方故障，并对这些计算中生成的数据提供高效的遗忘访问和搜索功能。研究小组将安全计算（MPC）和遗忘数据结构的先前工作技术结合起来，同时实现所有这些目标。该项目的影响是使新的隐私保护计算、分析和应用程序适应现实世界分布式系统的需求。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。

Secure multi-party computation (MPC) allows multiple distrusting parties to perform joint computations on their data without exposing the data to one another. Due to their promising security guarantees, the cryptographic protocols for MPC have gained unprecedented exposure in the last decade. With the proliferation of large, distributed systems and computations such as Tor and blockchain, there is an urgent need for large-scale MPC to ensure security and privacy for these applications. However, large-scale applications pose some unique challenges for the design of MPC protocols. There are thousands of parties, often with unequal resources, and some number of parties are likely to fail during a protocol execution. This project proposes to design MPC protocols to address these challenges. The project’s novelties are the design of new MPC protocols satisfying the requirements of large-scale distributed applications. Specifically, the investigators put forward a new framework for committee-based MPC for large-scale secure computations. These protocols distribute the computations across thousands of parallel committees to maximize the utilization of available resources for improved throughput. The designed protocols optimally choose committees and committee sub-protocols that optimize resource utilization and security. Additionally, they handle party failure and provide efficient oblivious access and search capabilities over data generated in these computations. The team of researchers combines techniques from prior work on secure computation (MPC) and oblivious data structures to simultaneously achieve all of these goals. The project’s impacts are to enable new classes of privacy-preserving computation, analytics, and applications geared to the needs of real-world distributed systems.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.