viden iden attac acker er ident dentif ifica ication ion
play

Viden: iden: Attac acker er Ident dentif ifica ication ion on - PowerPoint PPT Presentation

Viden: iden: Attac acker er Ident dentif ifica ication ion on on In- n- Vehic ehicle le Net Networ orks ks Kyong-Tak Cho and Kang G. Shin 1 Cont ontent ent Mo%va%on CAN Viden Evalua%on Drawback Future Work 2


  1. Viden: iden: Attac acker er Ident dentif ifica ication ion on on In- n- Vehic ehicle le Net Networ orks ks Kyong-Tak Cho and Kang G. Shin 1

  2. Cont ontent ent Ø Mo%va%on Ø CAN Ø Viden Ø Evalua%on Ø Drawback Ø Future Work 2

  3. Cont ontent ent Ø Mo%va%on Ø CAN Ø Viden Ø Evalua%on Ø Drawback Ø Future Work 3

  4. Mot otiv ivation ion ü Advancements in Automo%ve Technology ü Drawbacks in Present Defense Schemes 4

  5. Wha hat is is ECU? U? • Electronic Control Unit(ECU) • Types of ECU’s • Threats 5

  6. What is ECU? 6

  7. Related Work • Clock based Intrusion detec%on system • Clock Skews • Works only in Periodic message • APacker informa%on evaded –a periodic messages • Mean square voltage measurements • Works only with slow speed (10kbps) • Supervised Batch learning Algorithm • Not prac%cal 7

  8. VIDEN: Voltage based attacker IDENtification • Voltage measurements output by transmiPer ECU • Creates Voltage profiles (Fingerprints) based on voltage instance • Adap%ve signal processing( Online learning) • Defense mechanism against • Naïve adversary • Timing-aware adversary • Timing-voltage-aware adversary 8

  9. Content • Mo%va%on • CAN • Viden • Evalua%on • Drawback • Future Work 9

  10. Controller Area Network Protocol 10

  11. CAN typical application Schematic 11

  12. CAN output Voltage 12

  13. CAN Data Frame 13

  14. Content • Mo%va%on • CAN • Viden • Evalua%on • Drawback • Future Work 14

  15. System em and and Thr hrea eat model model CAN bus considera%on for system model • Fingerprin%ng device- IDS, %ming and voltage based • ECU aPached only through CAN bus Threats involved are naïve, %ming-aware and %ming-voltage-aware adversaries 15

  16. High-Le High-Level el Ov Over erview iew of of Viden iden Involves Four Phases • Phase 1: Learning ACK Threshold • Phase 2: Derives Voltage instances • Phase 3: Creates Voltage Profiles • Phase 4: Verifica%on 16

  17. Phas hase e 1: 1: Lear Learning ning ACK K Thr hres eshold hold • Measuring dominant voltages • Extrac%ng Non-ACK voltages 17

  18. Phas hase e 2: 2: Der eriv iving ing A Volt oltage ge Ins nstance ance 18

  19. Phas hase e 3: 3: Attac acker er Ident dentif ifica ication ion 19

  20. Phas hase e 4: 4: Ver erif ifica ication ion • Birthday paradox ü Voltage profile collision ü Mul%ple ECUs can have same profile ü Narrower set up of ECU to look at Target impersona%on ü Further verifica%on required to complement the Phase 1-3 20

  21. Content • Mo%va%on • CAN • Viden • Evalua%on • Drawback • Ques%on 21

  22. Evaluation • Against Timing adversary • Against Timing and Voltage adversary 22

  23. Content • Mo%va%on • CAN • Viden • Evalua%on • Drawback • Ques%on 23

  24. Drawbacks • APack from another network ECU • Atleast One Voltage profile • No message send from the ECU – Inaccurate iden%fica%on • Voltage profile adjustments 24

  25. Content • Mo%va%on • CAN • Viden • Evalua%on • Drawback • Ques%on 25

  26. Question 26

  27. THANK YOU 27

Recommend


More recommend