Ouri Wolfson's talk on using IT for Intelligent Transportation

This afternoon, I listened to Prof. Ouri Wolfson of UIC talk at the DAIS seminar about using IT in managing transportation in an intelligent manner. The talk title was Information Technology and Intelligent Transportation - A Marriage Made in Heaven.

Prof. Wolfson presented some futuristic ideas ... if vehicles could talk to each other, if things such as vehicle speed, road congestion, brake malfunction etc. could be exchanged between adjacent vehicles, and then propagated to a larger area, what things can we do? Wolfson argued that we could do a LOT. For example, if a road is congested, and our car learned that, we can take an alternate route. If suddenly a vehicle in front of a my car lost control, my car could take proactive steps to avoid an accident.

There are of course sensors showing traffic congestion, but those are usually deployed at main highways. Wolfson's vision is to create a vehicular p2p system that will work for arterial roads as well. There are of course many issues such as how cars exchange and propagate the info.

Prof. Wolfson showed a small box-device that can collect vehicle movement/location info, and then transmit that to other such devices. The vision is to have these devices deployed on a large number of cars, and then beam the aggregated system state to all of these devices. The device can then show a map, and provide situational awareness to the entire set of vehicles.

The work is part of the IGERT initiative, for intelligent transportation.


My take: As a security researcher, the first thing that comes into my mind is how we would handle security issues. Thinking like an attacker, I see a lot of ways to subvert the system. For example, how do we know that the boxes are not lying and feeding fake data? If I want other vehicles to erroneously believe a side road is congestion free, move over there, and hence clear the road ahead of me, shouldn't I (or a lot of people) be tempted to do that?

Besides lying for personal benefits, there may be more sinister reasons to cheat/subvert the system. Attackers can also cause congestion, and jam critical highways. Attackers can also feed fake "accident" information to cause nearby vehicles take emergency evasive action, which itself can cause more problems.

When we take information from a mass of potentially untrusted sources, we have to be careful from the outset. Unless we can ensure the provenance of information, and also verify trustworthiness of things we get fed from others, it is better not to look at the info.

People don't even have to be malicious ... a vehicular sensor unit can malfunction ... how do we ensure it won't cause a ripple effect and mess up the entire system?

---
I won't even touch on the "Privacy" issues i.e. Big brother like monitoring of cars ... though this is a big concern as well.

Of course, the networking and data propagation issues are quite interesting by their own right, and that's where Ouri Wolfson focused on the talk. Quite interesting.