1. Mini-Project 2007 On Location Privacy in Vehicular Mix-Networks Julien Freudiger IC-29 Self-Organised Wireless and Sensor Networks Tutors: Maxim Raya Márk Félegyházi

2. Mini-Project 2007 2 Outline 1.Problem Statement 2.System Model –Vehicular Networks –Adversary –Mix-zone 3.Cryptographic Mix-zones –The CMIX protocols 4.Vehicular Mix-Networks –Dynamic Mix-Networks 5.Results

3. Mini-Project 2007 3 1. Problem Statement What location privacy?

4. Mini-Project 2007 4 Our Approach Create Mix-zones Use Pseudonyms

5. Mini-Project 2007 5 2. Vehicular Networks Safety messages –position (p), speed (s) and acceleration (a) –Time stamp Assume Public Key Infrastructure (PKI) –Certification Authority (CA) distributes pseudonyms P i,k with k=1,…,F for vehicle i –To each P i k corresponds public/private key pair (K i,k,K i,k -1 ) P i k = H(K i,k )

6. Mini-Project 2007 6 Adversary Model Adversary types: 1.Weak Adversary (WA) Global Passive External with incomplete information 2.Strong Adversary (SA) Global Passive External with complete information 3.RSU Adversary –Global Passive partially Internal with complete information Local/Global: Monitoring area Internal/External: Member of the network or not Active/Passive: Alter information or not Complete/Incomplete Information: Amount of information

7. Mini-Project 2007 7 Mix-Zones definition Goal: Obscure relation of incoming and outgoing traffic => Unlinkability Strong adversary observes location and time of entering/exiting events: – Entering event: k = (n,  ) i.e., on road n at time  – Exiting event: l = (e,  ’)i.e., on road e at time  ’ Strong adversary has statistical information about mix-zones – Location: p n,e = Prob(“Vehicle enters on road n and exits on road e”) – Timing:q n,e (t) = Prob(“Time spent between n and e is t”) Pr k ! l = Prob(“ Mapping of entering event k to exiting event l ”)

8. Mini-Project 2007 8 Mix-Zones Effectiveness Measure effectiveness with entropy : Maximize entropy –High density (N) –High unpredictability (p,q) Mix-zones at road intersections where N= # of vehicles =>

9. Mini-Project 2007 9 3. Cryptographic Mix-Zone Silent Mix-zones: –Turn off transceivers –Unconditional security Cryptographic Mix-zones (CMIX): –Encrypt Safety Messages –Symmetric Cryptography –Computational security –Not user centric Not in the scope of Vehicular Networks

10. Mini-Project 2007 10 Centralized CMIX Protocol (p i,s i,a i ) = Safety message of vehicle i Ts = Time stamp Sign = Digital Signature Cert i,k = k-th Certificate of vehicle i SK= Symmetric Key

11. Mini-Project 2007 11 Distributed CMIX Protocol (p i,s i,a i ) = Safety message of vehicle i Ts = Time stamp Sign = Digital Signature Cert i,k = k-th Certificate of vehicle i SK= Symmetric Key

12. Mini-Project 2007 12 Centralized CMIX Protocol RSUs Adversary (p i,s i,a i ) = Safety message of vehicle i Ts = Time stamp Sign Ring = Ring Signature Desc Ring = Ring description SK= Symmetric Key Ring Signatures : Anonymous signatures based on groups Require public keys of all the group members Accountable signature scheme

13. Mini-Project 2007 13 4. Vehicular Mix-Networks Mix-network cumulative entropy for vehicle v: where L= Length of the path

14. Mini-Project 2007 14 Dynamic Mix-Networks Dynamics Set of traversed mix-zones always different Mix-zones have different q n,e (t) Path length L varies for each vehicle v –L v ~ N(  v,  v ) Upper Bounds WA model in Vehicular Mix-zone: –H(v) · log 2 (N) WA model in Vehicular Mix-network: –E[log 2 (N)] · log 2 (E[N])

15. Mini-Project 2007 15 5. Simulation Setup Network model 10X10 Manhattan network with 4 roads/intersection N ~ Poisson(   ~ Uniform[0,T] Uniform random walk, p n,e ~ U(1/4) q n,e ~ N(  n,e,  n,e ) Metrics Entropy Cumulative Entropy Intersection Mapping Success Ratio (SR) Vehicle Mapping Success Ratio (SR)

16. Mini-Project 2007 16 Mix-Zone Entropy

17. Mini-Project 2007 17 Mix-Zone SR

18. Mini-Project 2007 18 Mix-Networks Entropy

19. Mini-Project 2007 19 Mix-Networks SR

20. Mini-Project 2007 20 Results - Discussion Achievable anonymity depends on  –Traffic conditions determine location privacy Resistance to privacy degradation –Dynamic mix-networks offer good resistance –Dynamic mix-networks are strong when global uniformity local diversity

21. Mini-Project 2007 21 Future Work Results on VANET simulator –More realistic delay characteristics q n,e (t) and traffic patterns Extending towards user-centric location privacy –Cooperation for privacy –Cost of privacy Ring signatures –Anonymous signatures scheme for mobile networks with non-repudiation

22. Mini-Project 2007 22 Conclusion Location privacy in vehicular networks Cryptographic mix-zones (CMIX) Dynamic mix-networks Bounds on anonymity High location privacy for various types of adversaries

23. Mini-Project 2007 23 Related Work A. R. Beresford. Mix-zones: User privacy in location-aware services. PerSec 2004 L. Huang, K. Matsuura, H. Yamane, and K. Sezaki. Silent cascade: Enhancing location privacy without communication QoS degradation. SPC 2005 M. Li, K. Sampigethaya, L. Huang, and R. Poovendran. Swing & Swap: User-centric Approaches Towards Maximizing Location Privacy. WPES 2006 R. Rivest, A. Shamir, and Y. Tauman. How to leak a secret. ASIACRYPT 2001

24. Mini-Project 2007 24 CMIX Discussion Extended mix-zone Overlapping mix-zones –Same SK over several mix-zones Attacks –As secure as symmetric crypto –Key establishement