Modelling Privacy for Off-line RFID Systems Flavio Garcia Radboud University Nijmegen together with Peter van Rossum RFIDSec 2009
Outline Current RFID privacy models A new model for off-line RFID systems that considers reader corruption Forward and self-stabilizing backwards privacy Protocols Conclusions
RFID Systems
Current RFID Models Permanent secure connexion Juels and Weis (2006) Vaudenay (2007) Avoine (2005) Fwd-Privacy
SafeUn-Safe Time
Narrow-FWD Private protocol [OSK03]
Many real systems are more complex Periodic connexion What kind of security can still be guaranteed? More information on the readers
Consider off-line systems where readers can be compromised
An adversary is a PPTA with access to the set of oracles O: CreateReader(R) CreateTag(T) Launch(R) Send(m,A) Result() CorruptTag(T) Sync() O+ = O {DestroyReader(R)}
Fwd and Bwd-Privacy Safe Un-Safe Unachievable! (Unless extra assumptions are made) Safe
Forward privacy
Self-stabilizing backwards privacy
Forward and Self-stabilizing Backwards Private Protocol (idea) new day! BO K ← h(k’+1) K’ ← h(k’) K ← h(k) MAC using k’ K to `talk’ with the reader K’ to `talk’ with the BO
Forward and Self-stabilizing Backwards Private Protocol
Previous protocol is vulnerable to de- synchronization attacks Problem
Forward and Self-stabilizing Backwards Private Protocol
Verify key update Improvement
Improving synchronization
But still de-syncs if a reader is compromised Almost there
Improving synchronization
What to do Take special measures when a reader is compromised. Only update k’’s in BO if no reader corruption Con: this extends the privacy lost by one time slot
Conclusions model for (off-line) RFID systems in the presence of reader corruption forward and self-stabilizing backwards private protocols that uses only hash functions. De-sync resilience