Acumen: A Platform for Encrypted and Accountable Collaborative Editing
Ryan Cottone, Darya Kaviani, Conor Power, Will Giorza, Evelyn Koo, Natacha Crooks, Raluca Popa
摘要
Modern-day collaborative editing tools must reconcile a prominent tension between user privacy and collaboration: encrypting user data prevents an application server from processing user edits. We present Acumen, a cryptographic system for real-time collaborative applications based on conflict-free replicated data types (CRDTs). Acumen is the first system to provide strong snapshot consistency, enabling untrusted users to create verifiable document snapshots used to invite new collaborators. Acumen also provides confidentiality, integrity, fork-causal consistency, and ensures that invited users do not learn the previous edit history of the document. We achieve these properties through the use of cryptographic accumulators and a novel secure garbage collection mechanism. Our evaluation shows that Acumen can support 25 users each simultaneously typing 60 WPM with negligible degradation in latency and availability.
Existing secure decentralized editors fail to provide all of these guarantees. For instance, the state-of-the-art system Snapdoc [24] leaks user access patterns (violating confidentiality), provides a weak notion of snapshot consistency (violating secure dynamic membership), and has performance scaling with the overall edit history length.
Acumen is the first secure collaborative editor to provide confidentiality, integrity, and secure dynamic membership against malicious users and network adversaries. We also provide real-time performance: Acumen is able to sustain 25 users each typing at 60 WPM, orders of magnitude greater than the state-of-the-art. Acumen's performance and guarantees hinge on two key ideas:
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了最后一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
它引用的顶会 Paper2
相关 Paper
- Composing CRDTs Convergent by ConstructionAlexander Städing Dominguez, George Zakhour, Pascal Weisenburger, Guido SalvaneschiOOPSLA 2026
- Data-strophy: When Your Integrity Goes Wild, So Does Your Data!Ya-Nan Li, Yaqing Song, Qiang Tang, Moti Yung 等CCS 2026
- End-to-End Encrypted Collaborative DocumentsChristian Knabenhans, Zayd Maradni, Carmela TroncosoUSENIX Security 2026
- Keep CALM and CRDT OnShadaj Laddad, Conor Power, Mae Milano, Alvin Cheung 等VLDB 2023 · 被引用 13 次
- Abstraction for conflict-free replicated data typesHongjin Liang, Xinyu FengPLDI 2021 · 被引用 9 次
