Data-Dependent Confidentiality in DCR Graphs
summary
DCRSec implements a runtime information-flow monitor over Dynamic Condition Response graphs, uses value-dependent labels and a powerset lattice to give per-instance confidentiality, and enforces both explicit and two kinds of implicit leaks (marking changes and secret-enabled public events). The system hooks to an API, prevents disallowed reads and event executions, and the authors prove termination-insensitive non-interference, soundness and trace transparency for the monitor — a concrete enforcement mechanism, not just an analysis.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:11Dynamic Condition Response graphs
- 7:52Core relations in DCR
- 11:26Security compartments and access control
- 13:52Value-dependent information flow control
- 15:16Implicit information flow leaks
- 18:53Dynamic information flow monitor
- 20:18Non-interference in DCR graphs
- 21:02Privacy in business processes
- 22:08Conclusion and future research
- 22:55Q&A
related talks
13mDCR Graphs as Co-created Executable Models of the Law
30mSharing without Sharing: the Case for Data Clean Rooms Powered by Confidential Computing
53mI don't need privacy, I got confidentiality!
01h 08mTutorial: Security
47mThe Rise of Confidential Computing, What It Is and What it Means to You
20mData Race Freedom à la Mode
17mBreaking the Lethal Trifecta: Secure Agentic Computing via...
14mSecRSL: Security Separation Logic for C11 Release-Acquire Concurrency