Transportation Hack 2.0: Flaw Found Calif.'s FasTrak System • Hacker News Board

Re: Transportation Hack 2.0: Flaw Found Calif.'s FasTrak System

#11

Post by gojomo » Wed, Aug 27, 2008, 10:47 PM UTC

tptacek wrote: post #8

maximilian wrote: post #6

Earlier quoted context omitted.

A lot of controllers have builtin encryption hardware. If you put encryption in the original spec, its not that hard to include.

I can see it being easy to add encryption to a 32 bit part deployed in a pizza box form factor on the bottom of a telephone pole or alongside a train track right of way. I'm not sure it's that easy when you're constrained to a 16 bit part that can't big bigger than a wallet, has almost no power available, needs to be distributed in volumes of hundreds of thousands or more, needs to cost almost nothing per part, and n…

Until there's evidence someone is actually exploiting this at an economically relevant level, I wouldn't agree the "system needs to be more secure".

And if it is being exploited, using the existing license-plate cameras may be a more effective means of securing the system than upgrading the transponder behavior.

Re: Transportation Hack 2.0: Flaw Found Calif.'s FasTrak System

#12

Post by tptacek » Thu, Aug 28, 2008, 12:28 AM UTC

gojomo wrote: post #11

tptacek wrote: post #8

Earlier quoted context omitted.

I can see it being easy to add encryption to a 32 bit part deployed in a pizza box form factor on the bottom of a telephone pole or alongside a train track right of way. I'm not sure it's that easy when you're constrained to a 16 bit part that can't big bigger than a wallet, has almost no power available, needs to be distributed in volumes of hundreds of thousands or more, needs to cost almost nothing per part, and n…

Until there's evidence someone is actually exploiting this at an economically relevant level, I wouldn't agree the "system needs to be more secure". And if it is being exploited, using the existing license-plate cameras may be a more effective means of securing the system than upgrading the transponder behavior.

I think the first point is sensible, even though I disagree with it.

I'm not sure I understand how license plate cameras solve the problem, though. How many tens of thousands of license plates would need to be processed per day, and for what pattern?

Also note that the economics are just part of the problem. Other problems include privacy, chain of evidence, and personal fraud damages.

Re: Transportation Hack 2.0: Flaw Found Calif.'s FasTrak System

#13

Post by hhm » Thu, Aug 28, 2008, 2:43 AM UTC

tptacek wrote: post #10

hhm wrote: post #9

Earlier quoted context omitted.

I didn't say ROT13, it could be ROTx, or it could have a variable x (as with a big key), whatever... the advantage would be to make it slightly more difficult to find the meaning of the numbers in the card. I know security by obscurity doesn't work, but it's better than nothing.

I think it would have been even worse for FasTrak if Nate had got up on stage and said that they tried to pass XOR off as encryption. Then the story would not only be negligence, but also incompetence. There's a plausible story you can tell for why FasTrak isn't secured at all.

You are right.

Re: Transportation Hack 2.0: Flaw Found Calif.'s FasTrak System

#14

Post by gojomo » Thu, Aug 28, 2008, 5:45 AM UTC

tptacek wrote: post #12

gojomo wrote: post #11

Earlier quoted context omitted.

Until there's evidence someone is actually exploiting this at an economically relevant level, I wouldn't agree the "system needs to be more secure". And if it is being exploited, using the existing license-plate cameras may be a more effective means of securing the system than upgrading the transponder behavior.

I think the first point is sensible, even though I disagree with it. I'm not sure I understand how license plate cameras solve the problem, though. How many tens of thousands of license plates would need to be processed per day, and for what pattern? Also note that the economics are just part of the problem. Other problems include privacy, chain of evidence, and personal fraud damages.

They already take photos of, at the very least, those who slip through without paying or with a missing/faulty/dead transponder.

So step one is just to take a sample: for X thousand FasTrak transactions, for how many did the license plate and transponder disagree? If it's trivial, there's no economic rationale for fixing the system. Just make a note that transponder logs have a Y% fraud/error rate, for when disputes arise.

If still concerned about the other aspects, save the photo log for exactly as long as the transponder log. That reduces all the issues to the same as those with using counterfeit plates, which we've lived with for over a century, without major problems.

(In fact, I suspect they're already keeping the photo logs, but time will tell.)

Re: Transportation Hack 2.0: Flaw Found Calif.'s FasTrak System

#15

Post by tptacek » Thu, Aug 28, 2008, 7:07 PM UTC

gojomo wrote: post #14

tptacek wrote: post #12

Earlier quoted context omitted.

I think the first point is sensible, even though I disagree with it. I'm not sure I understand how license plate cameras solve the problem, though. How many tens of thousands of license plates would need to be processed per day, and for what pattern? Also note that the economics are just part of the problem. Other problems include privacy, chain of evidence, and personal fraud damages.

They already take photos of, at the very least, those who slip through without paying or with a missing/faulty/dead transponder. So step one is just to take a sample: for X thousand FasTrak transactions, for how many did the license plate and transponder disagree? If it's trivial, there's no economic rationale for fixing the system. Just make a note that transponder logs have a Y% fraud/error rate, for when disputes…

It's an interesting point. Obviously, if they had you working for the toll system, they'd be able to come up with some interesting countermeasures. Of course, they don't have anyone like you or I working there.