Establishing Cooperative Computation with Hardware Embassies

2022 IEEE International Symposium on Secure and Private Execution Environment Design (SEED)(2022)

引用 0|浏览3
暂无评分
摘要
While recent cryptographic techniques enable cooperative multi-party client-server computations under mutual distrust, they also introduce an efficiency tradeoff. Hosting all of the computation from the different parties involved on one set of servers requires everyone to agree on which servers are trustworthy. On the other hand, keeping the computations truly distributed introduces significant delays because of the inherently latency-sensitive nature of the protocols involved. In this paper, we explore the architectural impact of a possible middle path to this problem: resource-poor but physically secure devices interacting with significant (but not mutually trusted) compute and storage resources. The idea is that a small and well-protected "Embassy" can serve as a plot of sovereign soil in an otherwise untrusted environment. Building on techniques from multiparty computation (MPC) we show how such an architecture, even when extremely limited in size, can leverage local network capabilities and asymmetries in cryptographic operations to perform more efficient interactive secure computations. Even with a client-side device 5x slower, we show that common MPC applications can still be accelerated by 3x on average. Moreover, we explore the potential for architectural changes to further support multi-party evaluation through the addition of dedicated evaluator hardware further improving performance 1.52x.
更多
查看译文
关键词
hardware embassies,cryptographic techniques,mutual distrust,latency-sensitive nature,architectural impact,possible middle path,physically secure devices,storage resources,multiparty computation,suchan architecture,cryptographic operations toperform,potentialfor architectural changes,support multiparty,evaluator hardware,interactive secure computations,well-protected embassy,cooperative multiparty client-server computations,protocols,asymmetric cryptographic operations,client-side device,multiparty evaluation
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要