evading cellular data monitoring with human movement
play

Evading Cellular Data Monitoring With Human Movement Networks Adam - PowerPoint PPT Presentation

Evading Cellular Data Monitoring With Human Movement Networks Adam J. Aviv, Micah Sherr*, Matt Blaze, and Jonathan M. Smith University of Pennsylvania, *Georgetown University Adam J. Aviv University of Pennsylvania HotSec '10 1 Motivation


  1. Evading Cellular Data Monitoring With Human Movement Networks Adam J. Aviv, Micah Sherr*, Matt Blaze, and Jonathan M. Smith University of Pennsylvania, *Georgetown University Adam J. Aviv University of Pennsylvania HotSec '10 1

  2. Motivation Adam J. Aviv University of Pennsylvania HotSec '10 2

  3. Goal Out-of-Band communication Unmonitored and completely decentralized Adam J. Aviv University of Pennsylvania HotSec '10 3

  4. HumaNet Human-to-Human Mobile Ad-Hoc Network Humans + Smartphones Adam J. Aviv University of Pennsylvania HotSec '10 4

  5. HumaNet Adam J. Aviv University of Pennsylvania HotSec '10 5

  6. Design Trade-ofs Complete Random Epidemic HumaNet Centralization Walk Reliability Network Load Latency Anonymity Adam J. Aviv University of Pennsylvania HotSec '10 6

  7. Regularity of Movement Patterns Adam J. Aviv University of Pennsylvania HotSec '10 7

  8. Return-to-Home Principle A person is likely to return to places frequented in the past Adam J. Aviv University of Pennsylvania HotSec '10 8

  9. HumaNet Protocol Idea No further duplication of messages Address message to recipient's likely future locations Local routing decision based on movement history Adam J. Aviv University of Pennsylvania HotSec '10 9

  10. Alice and Bob ... ? B A Adam J. Aviv University of Pennsylvania HotSec '10 10

  11. A C Adam J. Aviv University of Pennsylvania HotSec '10 11

  12. D ` D Adam J. Aviv University of Pennsylvania HotSec '10 12

  13. B A D C Adam J. Aviv University of Pennsylvania HotSec '10 13

  14. Routing Refnements Local Timeout Global Timeout Last Mile Flooding Adam J. Aviv University of Pennsylvania HotSec '10 14

  15. Constructing a Profle One Day's GPS locations Adam J. Aviv University of Pennsylvania HotSec '10 15

  16. Cluster Points Adam J. Aviv University of Pennsylvania HotSec '10 16

  17. One Day's Homes Adam J. Aviv University of Pennsylvania HotSec '10 17

  18. Combine With Other Days Adam J. Aviv University of Pennsylvania HotSec '10 18

  19. Trace Driven Simulation Adam J. Aviv University of Pennsylvania HotSec '10 19

  20. Data Source Cabspotting Dataset 20 days, 536 Cabs in San Francisco Adam J. Aviv University of Pennsylvania HotSec '10 20

  21. Comparison Epidemic Flooding Probabilistic Flooding Random Walk Adam J. Aviv University of Pennsylvania HotSec '10 21

  22. #Messages Required Adam J. Aviv University of Pennsylvania HotSec '10 22

  23. Message Latency 76% w/in 1 day Adam J. Aviv University of Pennsylvania HotSec '10 23

  24. Successful Delivery Adam J. Aviv University of Pennsylvania HotSec '10 24

  25. Challenges Reliability Routing Attacks Location Privacy Anonymity Adam J. Aviv University of Pennsylvania HotSec '10 25

  26. Reliability and Routing Attacks Best-Efort routing How reliable would we need? Peer-to-Peer System Vulnerable to same class of attacks, but how feasible are they here? Adam J. Aviv University of Pennsylvania HotSec '10 26

  27. Location Privacy Periodic broadcasts of location information Peoples willingness to participate? Reveal surprising locations? Adam J. Aviv University of Pennsylvania HotSec '10 27

  28. Anonymity Can this system provide Anonymity? Sender Anonymity message timeout leaks info Receiver Anonymity Message no longer being passed Broadcast in crowds (k-anonymity) Adam J. Aviv University of Pennsylvania HotSec '10 28

  29. Brain Storming ... Attacking HumaNet, how would you do it? Necessary resources? Feasible? Would you participate? If not, what would you need to say yes? Adam J. Aviv University of Pennsylvania HotSec '10 29

  30. Thanks Questions, Discussion? Adam J. Aviv University of Pennsylvania HotSec '10 30

  31. Generate a Home Adam J. Aviv University of Pennsylvania HotSec '10 31

  32. How Predictive? Average: 65% of GPS coordinates fell within homes 65% of the day (time) Worst Case: 39% of GPS coordinates fell within homes 45% of the day (time) Adam J. Aviv University of Pennsylvania HotSec '10 32

  33. Other Routing Protocols Epidemic Pocket Switched Networks [CHCDGS'07] Pollen [GSM'01] Ad-Hoc DREAM [BCSW'98] GPSR [KK'00] Geographic Ad-Hoc'ish Adam J. Aviv University of Pennsylvania HotSec '10 33

Recommend


More recommend