Explore a conference talk on Ghostor, a secure data-sharing system built on decentralized trust. Learn about a novel approach to protecting user privacy and data integrity from compromised servers. Discover how Ghostor hides user identities and allows detection of server-side integrity violations without relying on centralized trust. Understand the system's innovative design, which avoids per-user authentication and user-specific mailboxes. Delve into the concept of verifiable anonymous history and how Ghostor utilizes blockchain technology minimally. Examine the system's architecture, implementation challenges, and performance metrics. Gain insights into the trade-offs between security, privacy, and system overhead in data-sharing applications.
Ghostor - Toward a Secure Data Sharing System from Decentralized Trust
Overview
Syllabus
Intro
Motivating Example: Medical Record System
Ghostor: Cryptographic Data Sharing System
E2EE Data Sharing vs. Ghostor's Anonymity
Verifiable Linearizability
Comparison to Existing Work
Bootstrapping Decentralized Trust
Strawman: Use a Blockchain
Ghostor's System Architecture
Verifiable History (Strawman)
Shared Capabilities
Additional Challenge: Concurrent Operations
This Technique Does Not Work for Writes
Ghostor Stack
Implementation
Server-Side Latency to PUT a 1 MiB Object
Total Latency
Conclusion
Taught by
USENIX
