1. Norfolk State University Department of Computer Science Colloquium
NOTICE: An Architecture for the Notification of Traffic Incidents and Congestion
Dr. Michele C. Weigle
Department of Computer Science
Old Dominion University
(Work done with Dr. Stephan Olariu and Gongjun Yan)
Norfolk State University
Department of Computer Science Colloquium
March 1, 2007

2. Motivation Give drivers advance warning of upcoming traffic congestion
EXIT 12
Exit while you
still can!
From trekearth.com
Give drivers advance warning of upcoming traffic congestion

3. Outline Overview of Vehicular Ad-Hoc Networks (VANETs)
Security Issues in VANETs
Our Approach: NOTICE
Simulations
Conclusions

4. Vehicular Ad-Hoc Networks VANETs
Traffic information
cars report their position and speed to surrounding cars
car may suggest an alternate route
Weather warnings
Collision warning
Platooning
Intersection Assistance
C1 speed 0
C2 speed 0
C3 speed 0 … c1 c2 c4 c3 c5
platooning - increase highway throughput, highway speeds at close spacings (6.5 m, 21 ft)
EXIT 12

5. VANETs Approaches V2V only (zero infrastructure, purely ad-hoc)
require no outside infrastructure or roadside devices
vehicles communicate with each other to determine traffic situation
V2V and V2I
requires some outside infrastructure, often in the form of roadside devices
infrastructure can provide aggregation/processing, encryption key distribution, access to larger network

6. VANETs V2V / V2I Architecture
From “The Security of Vehicular Ad Hoc Networks”, M. Raya and J.-P. Hubaux, SASN 2005

7. Security Issues Adversaries
Greedy Drivers
convince neighbors that congestion is ahead to clear roads
Snoops
driver profiling, tracking
Pranksters
hack things “just for fun”
Industrial Insiders
if mechanics are in charge of uploading software, they can load malicious programs
Malicious Attackers
terrorists, criminals with specific targets in mind
Bryan Parno and Adrian Perrig. Challenges in Securing Vehicular Networks, HotNets 2005.

8. Security Issues Attacks
Denial of Service (DoS)
overwhelm a vehicle’s resources or jam communication channels
Message Suppression
selectively drop messages, suppress congestion alerts
Fabrication
broadcast false information into network
Alteration
alter existing data, replaying earlier transmissions, disrupt voting mechanisms
Bryan Parno and Adrian Perrig. Challenges in Securing Vehicular Networks, HotNets 2005.

9. Security Issues Approaches
Digitally sign (encrypt with private key) each message sent by a vehicle
a vehicle is issued a certificate from an authority
certificate verifies vehicle’s public key used for decryption
disadvantage: allows tracking of vehicles
Pre-load many different anonymous key pairs and change keys at certain intervals
disadvantage: malicious user could use the keys to impersonate multiple vehicles
Key: Reliably associate a message with physical vehicle

10. Our Approach NOTICE
Allow the roadway to associate messages with physical vehicles
Embed intelligent sensor belts in the highway
When a car passes over the belt, it reports its speed to the belt
The belt makes decisions about where congestion is occurring based on reports from cars and other belts
speed 55

11. NOTICE Car Model and Belt Model
EDR
BTd Td Th
BTh
Event Data Recorder (EDR)
tamper-proof
records location, speed, etc.
Two transceivers
one for handshaking, Th
one for data transfer, Td
Pressure sensors
detect passing cars
Two transceivers
one for handshaking, BTh
short range (~1m)
one for data transfer, BTd
larger range (~3m)

12. NOTICE Belt-to-Belt Communication
Individual belt in each lane
Connected belts (sub-belts) communicate instantaneously
Non-connected belts do not directly communicate
use cars as data mules
Gives encrypted message to a car to drop off at next belt
[avg spd 55]
[avg spd 55]

13. NOTICE Belt-to-Car Communication - HandshakingTh Td
EDR
BTh
Belt sends “Hello” message to car
ID of belt
frequency channel for further communication, 
one-time shared encryption key, 
Car sends short acknowledgement

14. NOTICE Belt-to-Car Communication - Data TransferTh Td
EDR 3m
BTd
Belt sends query
Car sends message from previous belt
Car sends encrypted (with ) EDR data
Belt sends encrypted (with ) traffic information
Belt sends encrypted message for next belt

15. NOTICE Information Propagation
B1 is aware of traffic slowdown
creates encrypted message with latest traffic statistics
Information is provided to B2
B2 uploads message onto car destined for C2
When C2 receives message, it provides it to C1
C1 notifies passing cars B2 B1
EXIT 12 C2 C1

16. NOTICE Urgent Mode A2 A1
B2 uploads message with urgent bit set onto car destined for C2
Car broadcasts message to other cars for faster delivery
Cars are passing encrypted messages, so no security risk B2 B1
EXIT 12 C2 C1

17. NOTICE Role-Based Communication
Emergency responders can provide information to NOTICE belts
Special encryption key used
Validate incident/congestion inference made by belts
EXIT 12

18. NOTICE Evacuations Evacuees need information about resources
gasoline, hotels, shelters, etc.
Emergency management centers need method to disseminate information
Enhanced NOTICE can provide this
temporary infrastructure connected to belts for long-range communication
to emergency management centers
for backward propagation during contraflow

19. NOTICE Evacuations Cars that have refueled report to nearest belt
Location and time of refuel propagated backwards by temporary infrastructure
Cars needing gas can exit at the appropriate location

20. NOTICE Simulations Developed a Java-based simulator
based on applet using realistic highway traffic model
Measured message propagation time
normal mode
car receiving message carries it to the next belt
urgent mode
car receiving message broadcasts it to nearby cars
Traffic intensities from 70 vehicles/hr to 3600 vehicles/hr

21. NOTICE Simulations

22. NOTICE Simulations

23. Conclusion
NOTICE: An Architecture for the Notification of Traffic Incidents and Congestion
Provides security and privacy
belts can independently corroborate information provided by vehicles
Works in sparse or dense traffic
Extensions for evacuation scenarios

24. Future Work Enhance our simulator
wireless channel conditions
Rules for how far to propagate congestion notifications
Rules for how to infer occurrence of traffic incident or congestion
Use with non-intrusive sensor

25. VANET Research Group @ ODU http://www.cs.odu.edu/~vanet
Michele C. Weigle
Department of Computer Science
Old Dominion University
Norfolk, VA
VANET Research ODU