1. Geoencryption Using Loran Di Qiu, Sherman Lo, Per Enge Stanford University Sponsored by FAA Loran Program

2. 2/28/20162007 Convention and Technical Symposium - ILA-362 Why Geoencryption? Unsecure world  Data/Information security  Piracy concern Traditional cryptosystems have inconveniences or weaknesses  Something you know: PIN, passwords  Something you have: key, smart card  Something you are: biometrics

3. Location for Security  Universality Do all people have it?  Collectability How well can an identifier be captured or quantified?  Circumvention foolproof  Uniqueness Can people be distinguished based on an identifier? 2/28/20162007 Convention and Technical Symposium - ILA-363

4. 2/28/20162007 Convention and Technical Symposium - ILA-364 ----------- ----------- ----------- ----------- ------- Encryption and Authentication ----------- ----------- ----------- ----------- ------- @#)&)* +!#$N$ &*&$& =W& /- !&)$#+ Encryption Communication Channel @#)&)* +!#$N$ &*&$& =W& /- !&)$#+ ----------- ----------- ----------- ----------- ------- Authentication ----------- ----------- ----------- ----------- ------- Communication Channel ----------- ----------- ----------- ----------- ------- Verify

5. Geoencryption 2/28/20162007 Convention and Technical Symposium - ILA-365 ----------- ----------- ----------- ----------- ------- @#)&)* +!#$N$ &*&$& =W& /- !&)$#+ Sender Receiver geotag

6. Geodecryption 2/28/20162007 Convention and Technical Symposium - ILA-366 ----------- ----------- ----------- ----------- ------- Loran Receiver Feature Extraction @#)&)* +!#$N$ &*&$& =W& /- !&)$#+ Verify yes

7. Why Loran? GPS Non-stationary satellites High absolute accuracy, high repeatable accuracy Global coverage Low SNR – Easy to jam and spoof – Indoor NOT capable Data channel Loran Stationary transmitters Low absolute accuracy, high repeatable accuracy Northern hemisphere High SNR – Hard to jam and spoof – Indoor capable Data channel (eLoran) 2/28/20162007 Convention and Technical Symposium - ILA-367

8. Security Analysis Outline 2/28/20162007 Convention and Technical Symposium - ILA-368 Security Vulnerabilities of Protocol/Implementation I. Spoof II. Replay III. “Parking Lot” Attack Tag Size IV. Spatial decorrelation Loran Receiver Feature Extraction Loran Receiver Feature Extraction d Same tag? Modify location parameters

9. Signal Authentication  TESLA – Timed Efficient Stream Loss-tolerant Authentication  Authenticating message = key (K) + tag (h)  Tag = MAC (Data, Key) 2/28/20162007 Convention and Technical Symposium - ILA-369 MAC h’ ?= h ELR mhKmhKmhK I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

10. Authentication Test Middletown Circular TESLA chain 50% Bandwidth Message subtypes – Type 1-4 (0001-0100): first 148 bits of the tag – Type 5 (0101): last 12 bits of tag, – Type 6-9 (0110-1001): first 148 bits of key – Type 10 (1010): last 12 bits of key 2/28/20162007 Convention and Technical Symposium - ILA-3610 00100011010001010111100010011010 tagkey 00010110 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

11. Authentication Test Result 2/28/20162007 Convention and Technical Symposium - ILA-3611 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

12. Loran Certified Receiver 2/28/20162007 Convention and Technical Symposium - ILA-3612 Navigation Receiver Signal Authentication Module Geotag Generation Module Location Verification Decryption Original file Tamper Resistant Loran input @#)&)* +!#$N$ &*&$& =W& /- !&)$#+ File input I. Spoof II. Replay III. Parking lot IV.Spatial decorr. Replay Modified Location Information

13. Parking Lot Attack 2/28/20162007 Convention and Technical Symposium - ILA-3613  False Accept Rate (FAR): % of unauthorized persons accepted in error  False Reject Rate (FRR): % of authorized persons who are incorrectly denied acceptance  Trade off between FAR and FRR I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

14. Data Collection 2/28/20162007 Convention and Technical Symposium - ILA-3614 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

15. FAR & FRR Estimation 2/28/20162007 Convention and Technical Symposium - ILA-3615 Experimental Analysis Grid interval size Analytic Analysis userattacker FRR FAR Grid Size FAR = Q(interval size, , distance) FRR = Q(interval size,  ) I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

16. Receiver Operating Curve 2/28/20162007 Convention and Technical Symposium - ILA-3616 Grid Size Better Performance I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

17. Spatial Decorrelation  High spatial decorrelation is preferred.  Evaluation functions  Distance measure  Error rates measure - FAR  Information measure - relative entropy D(p||q)  Dependence measure - correlation coefficient 2/28/20162007 Convention and Technical Symposium - ILA-3617 distance high low I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

18. Test Locations 2/28/20162007 Convention and Technical Symposium - ILA-3618 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

19. False Accept Rate - Different Stations 2/28/20162007 Convention and Technical Symposium - ILA-3619 StationSNR (dB) Fallon21 George6 Middletown32 Searchlight8 High SNR results in high spatial decorrelation. I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

20. Decorrelation Distance FAR < 0.01 2/28/20162007 Convention and Technical Symposium - ILA-3620 Decorrelation distance is 18 meters for Middletown. I. Spoof II. Replay III. Parking lot IV.Spatial decorr. e -0.2445d

21. False Accept Rate - Different Location Parameters 2/28/20162007 Convention and Technical Symposium - ILA-3621 I. Spoof II. Replay III. Parking lot IV.Spatial decorr. TDOA/TOA > ECD > SNR

22. Geotag Size 2/28/20162007 Convention and Technical Symposium - ILA-3622 ParameterEntropy (bits) TDOA15.5 ECD6.0 SNR4.3 25.8 bits I. Spoof II. Replay III. Parking lot IV.Spatial decorr. 811 km  Information measure  Entropy  Key/Geotag size  Station coverage  Information density  Spatial decorrelation

23. Conclusion  Defeated vulnerabilities of geoencryption protocol and implementation  Signal authentication & certified receiver  Spatial decorrelation of Loran location parameters  Geotag size from Middletown is 26 bits  At least 2 26 trials of different locations to break it  How to increase geotag size?  Look into more parameters  Fuzzy extractor 2/28/20162007 Convention and Technical Symposium - ILA-3623

24. Acknowledgement The authors would like to thank Ben Peterson, Kirk Montgomery, Jim Shima and USCG for their help during the research. 2/28/20162007 Convention and Technical Symposium - ILA-3624

25. Backup Slides 2/28/20162007 Convention and Technical Symposium - ILA-3625

26. Demodulation Performance 2/28/20162007 Convention and Technical Symposium - ILA-3626 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

27. Message Loss 2/28/20162007 Convention and Technical Symposium - ILA-3627 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

28. Authentication Performance SNR Bandwidth 2/28/20162007 Convention and Technical Symposium - ILA-3628 I. Spoof II. Replay III. Parking lot IV.Spatial decorr. TESLA Segment Data Authentication (320 bits) 320/37  9 Loran messages 50% BW  18 Loran messages Authentication probability is proportional to SNR & BW.

29. TDOA Data 2/28/20162007 Convention and Technical Symposium - ILA-3629 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

30. Distribution of Quantized TDOA - Grid Size 20m, Station George 2/28/20162007 Convention and Technical Symposium - ILA-3630

31. PDF 2/28/20162007 Convention and Technical Symposium - ILA-3631

32. Receiver Operating Curve 2/28/20162007 Convention and Technical Symposium - ILA-3632 Grid Size Better Performance

33. TDOA Measurements 2/28/20162007 Convention and Technical Symposium - ILA-3633 I. Spoof II. Replay III. Parking lot IV.Spatial decorr.

34. Decorrelation Distances - Different Parameters 2/28/20162007 Convention and Technical Symposium - ILA-3634

35. Relative Entropies 2/28/20162007 Convention and Technical Symposium - ILA-3635