Modelling Memory for Individual Re-identification in Decentralised Mobile Contact Tracing ...
summary
Paper demonstrates a practical re-identification attack on decentralized BLE contact tracing by modeling human memory of people and places as a graph model with time windows and pruning edges where remembered counts match received ephemeral codes. Evaluation on Copenhagen (with RSSI) and Social Evolution datasets yields >90% identification in some scenarios, shows fake-contact injection fails, and finds two effective mitigations: variable time window / reduced data length and adjusting contact range by sociability (trades privacy for precision).
AI-written summary · about
related talks
01h 15mEXPOSURE NOTIFICATION DURING THE COVID-19 PANDEMIC: RECONCILING FUNDAMENTAL RIGHTS AND PUBLIC...
39mPrevent Phishing & Impersonation
14mHuman-Centered Contact Tracing Apps
46mFrom Health Data to Location Traces: Addressing Spatiotemporal Correlations in Differential Privacy
20mChameleon Devices: Investigating More Secure and Discreet Mobile Interactions via Active ...
01h 43mPerspective
13mIs it Worth Giving Up Privacy for Public Health Benefits?
06mCreepy stuff with anonymous data