Hacker News Reader

Comment by NelsonMinar | Hacker News Reader

The risk these things pose to aviation is real. There's been a huge shift towards GPS for instrument flying. Cross country flying frequently relies on GPS to know where you are. Even more important, flying an approach to land now often uses GPS to keep the plane to a specific safe path, essential when flying in a cloud. A plane isn't going to fly out of the sky if GPS suddenly stops working, but it can make life awfully difficult.

The risk is not just theoretical. For example, in 2009 the Newark airport had a daily GPS failure that was finally traced to a trucker using a jammer, presumably to defeat a tracker enforcing safe driving rules. http://www.economist.com/node/18304246

View on HN

The more sophisticated jammers don't cause "noise", but rather they record and replay the pseudorandom generated digits they receive from the various frequencies and time drift them. It's hard to triangulate much of anything then and it could send you off miles from where you think you are.

Much better than just noise because you wouldn't know there is a problem.

This attack is what Iran says they did to capture the US drone a few months ago. It's pretty crazy that GPS signals include no cryptographic authentication, even for military use.

I'm not so worried about this kind of attack on civilian aviation. It would be a directed attack: potentially deadly but rare and focussed. I'm much more worried about the casual disruption to GPS from truck drivers spending $99 on a grey market device.

There is cryptographic hash so that you can't send false data, but replaying the existing data but time delayed is still possible. The timing is how the triangulation fix is done. More sources of data, the harder it is to spoof. If you only have 3 or 4 fixed points, it's easy to spoof. 6 to 10 and it becomes extremely difficult.

Also military GPS is far more accurate than civilian GPS by a factor of 10x.

"military GPS is far more accurate than civilian GPS" - is this still true? I don't know how my iPhone GPS triangulates me, but it usually has me located to within 3-5' at all times when I'm walking home.

yep. the reason why your iPhone is fast has to do more with AGPS and that it can use wifi access points as references points as well.

I believe that difference went away in 2000: http://en.wikipedia.org/wiki/Selective_Availability#Selectiv...

Military hardware can access encrypted signals that provide greater precision:

http://en.wikipedia.org/wiki/Global_Positioning_System#Satel...

That has nothing to do with why the iphone is so fast.

Civilian transmission rates on L1 are 10x slower than the military L2 frequency. It takes a full 6 seconds to get a single frame downloaded and that is also hoping your almanac is up to date. (Thankfully AGPS can provide the almanac data by other menas.)

The military version it takes .6 seconds to do the same it takes us civilians to do in 6 seconds. The SA drift is gone now but still the slower speed makes fixing a perfect position slightly more difficult and less accurate.

I'm speaking out of ignorance here, but why couldn't you just use a quantitized time value (to account for transmission lag) as a part of the data that's hashed? It works great to prevent replay attacks with OAuth.

nope. it's time that is the factor for triangulation. you always take the first time you received from the one source and you can discard certainly something way out side of what you expect. It's how you defeat the concreate canyon effect in big cities where signals are bouncing off everything.

However if you have a low strength signal from one source and you receive a frame that is a second or a two off, it can cause you to think you are miles away from where you think you supposed to be.

I just saw a demonstration a few days a go by the UT Austin Radio Navigation lab where they spoofed L1 GPS and caused a radio controlled helicopter to somewhat precisely veer off course. I'm interested in the impact of this on aviation as well.

Reply on news.ycombinator.com