from the the-price-of-cheaper-insurance? dept.
Researchers have found that it is possible to determine where someone has driven to from just a starting point and their speed, without the need for GPS data.
A team of Rutgers University computer engineers has shown that even without a GPS device or other location-sensing technology, a driver could reveal where he or she traveled with no more information than a starting location and a steady stream of data that shows how fast the person is driving.
Insurance companies and customers both have incentive to monitor driving speeds, said Janne Lindqvist, assistant professor in the Department of Electrical and Computer Engineering at Rutgers. Drivers who avoid jackrabbit starts and sudden stops are typically lower-risk drivers, and insurance companies benefit by rewarding such behavior. So some companies are offering lower premiums to customers who install a device that constantly measures, records and reports their speed.
"The companies claim this doesn't compromise privacy, because all they are collecting is your speed, not your location," said Lindqvist, who is also a member of the university's Wireless Information Network Laboratory, or WINLAB. "But we've shown that speed data and a starting point are all we need to roughly identify where you have driven."
(Score: 1, Funny) by Anonymous Coward on Monday August 11 2014, @09:59PM
This is my favorite kind of discovery. One that I would never have come up with on my own but is obvious in hindsight.
(Score: 4, Insightful) by bob_super on Monday August 11 2014, @10:00PM
That took a team? I'm pretty sure any physics or math teacher could have pointed that out.
The only exception I could find to this being obvious would be the US grid-shaped cities. And even then, just hitting a couple major roads would dramatically shrink uncertainty. Add some timestamps to correlate to known bad traffic, and you're pretty sure to locate anyone after a few miles.
(Score: 0) by Anonymous Coward on Monday August 11 2014, @11:20PM
Used to write truck tracking software. Think one of my old bosses has a patent on this.
(Score: 0) by Anonymous Coward on Monday August 11 2014, @10:05PM
You fuckwits actually BELIEVE that the tracking device will only
retain speed data ?
I've got a bride to sell you. On second thought, I'd rather you all jump
off that bridge, and the sooner the better.
(Score: 5, Funny) by DECbot on Monday August 11 2014, @10:23PM
Honestly, I'm more interested in the bride you're selling than the bridge jumping, sir.
cats~$ sudo chown -R us /home/base
(Score: 1, Funny) by Anonymous Coward on Monday August 11 2014, @10:56PM
You must be outside the United States if you can purchase a bride. Here we can only lease them on an agreement which they can cancel at any time but it is us that pays the termination fees which exceed half our known, and sometimes even supposed, wealth. Continuous additional recurring costs can also be added depending on state and/or whether any minors are involved and she chooses to take them with her. These being only surface issues and you should consult a specialist in family law + CPA for specifics.
Now before any of you say this is offtopic, see the subject line and note that I am just replying to the thread...
(Score: 2) by kaszz on Tuesday August 12 2014, @03:19AM
Demand sterilization and prenuptial agreement ..?
(Score: 2, Funny) by Paradise Pete on Tuesday August 12 2014, @02:41AM
And if you knew my starting point and speed perhaps you'd be kind enough to dredge out my body afterward.
(Score: 3, Funny) by Anonymous Coward on Monday August 11 2014, @10:30PM
Now we can know speed and position.
(Score: 2) by FatPhil on Monday August 11 2014, @10:51PM
In that case, the most they need is absolute acceleration data (which would need to be extrapolated/integrated twice - which adds a lot more noise into their dead-reckoning input), but what they seem to be most interested in for classifying drivers is jerk, the third derivative. And they're not interested in it if it's below a threshold. I'm pretty sure that just a binary stream of
| d^3 x/dt^3 | > thresh
or if not binary a very low resolution quantized value, would tell them everything they need, for their claimed use. That would be unimaginably more noise for those who want your position. I'm not saying it's impossible, but it would be an actually difficult challenge rather than the seemingly rather trivial thing that they have done. Dead-reckoning from an accelerometer is pretty simple, and speed data is about as different from position data as accelleration data is. Losing the direction is a bummer, but you've got one less integration of the noise, two less integrations than my jerk proposal. Having the road map as an input is an enormous help.
Great minds discuss ideas; average minds discuss events; small minds discuss people; the smallest discuss themselves
(Score: 0) by Anonymous Coward on Monday August 11 2014, @11:07PM
There are times too where the accelerator and proper steering can get you out of trouble where the brake can get you dead, particularly involving the errors of others.
(Score: 0) by Anonymous Coward on Tuesday August 12 2014, @02:34PM
(Score: 2) by Snotnose on Tuesday August 12 2014, @12:16AM
On the freeway I match traffic, passing few and being passed by a few. Which here in San Diego means I'm doing 75-80 MPH. My insurance company would be all over me for that.
As for jackrabbit starts, hard braking, and (unmentioned in the article) hard cornering, I quit doing that when I hit 40 or so and it just wasn't worth the gas/tire/brake expenses anymore.
When the dust settled America realized it was saved by a porn star.
(Score: 3, Interesting) by bob_super on Tuesday August 12 2014, @12:30AM
I'm pretty sure that lateral acceleration is the only thing that insurances would really need.
They really need to catch the people doing something else and drifting back and forth in their lane. That's a better indicator of dangerosity than limited speeding.
(Score: 0) by Anonymous Coward on Tuesday August 12 2014, @12:55PM
(Score: 1) by tftp on Tuesday August 12 2014, @12:50AM
It is good that reliability of such a method is not extremely high. One can be stuck in traffic on a 65 mph freeway, or can exceed the posted speed on an empty 45 mph surface street. The method completely fails if all streets in your area offer the same speed of travel, same size of city blocks, or few controlled intersections.
But what's really bad is that consumers of such data do not care about accuracy. If an insurance company gets hold of this data, they will use it to increase your risk factor no matter what you do. If the police obtains this data, they will add you to the list of suspects if the liberally interpreted data places you anywhere within 10 miles from the crime scene. It will be your responsibility then to prove that you are not guilty, as the prosecution already has all that they need - the calculations that clearly show to the jury that you did it.
(Score: 2) by Bot on Tuesday August 12 2014, @06:39AM
Account abandoned.
(Score: 1) by tftp on Tuesday August 12 2014, @07:17AM
As somebody pointed out, obvious, in hindsight.
It's obvious... but it was never mentioned before just because it is a poor man's way of dead reckoning. For example, the start location may be not known. (How would, say, an insurer get that location without a pair of absolute coordinates?) In practicality, if the said insurer can only access the log twice per year, when you come to their office (who does that?) they can only have one known location for 6 months. Everything else has to be calculated.
Then we have traffic problems - not only slowdowns, but also detours. They may be not on the maps. Combined with everything else, like a jaywalker, for who you had to brake in the middle of a 45 mph highway with no traffic lights, it becomes a complex problem of correlation. You have a map of the city with all the traffic lights, their timing, and all the distances. You have a fuzzy set of measurements. You have to find a graph that optimally fits the map - if you only drive on the map, and not in private parking lots (that can be huge) and not outside of the city (this method does not work there at all.) You will end up with a solution... or with a thousand solutions, and the difference between them would be significant, whereas their "quality" (that represents the correlation) would be about the same.
To illustrate the later. As you know, many US cities are built on a rectangular grid. You can take many different paths on this grid from A to B. All these paths will have a similar number of turns, and the same speed - but they do not let you know where you went. Take a 2x2 map. You start in the (-1,-1) corner. The speed readings tell you that you took three edges of that graph (each edge is 1 unit.) Where did you end up? Well, almost anywhere! You could even make a U turn and go back, and more than once! This method may work in old cities of Europe that are highly irregular, and where each segment between two traffic lights is identifiable. But it may not be so great in US cities that have blocks of similar size and a rectangular grid for streets, and every reason to take one street today and another street tomorrow.
All in all, this is a very inefficient method of tracking. This is why it was not proposed before. There are many things that are still not proposed for the same reason - they are foolish things to do. If the car's position is worth of being known, it should be recorded reliably - or not at all.
(Score: 2) by marcello_dl on Tuesday August 12 2014, @12:39PM
I'd say the usage patterns of cars let you guess their position more often than once in 6 months.
Example:
- car shut down for 8 hours?: sleep at home
- car doing infrequent 40 minutes stop after work and then home in a hurry: mistress
(Score: 2) by tangomargarine on Tuesday August 12 2014, @02:31PM
Unless of course you decide to double back on your route to screw up the math. What this gives you is a fuzzy circle of how far you could have traveled. After all, it's not impossible that you were driving 30mph on a 65mph freeway, just rather unlikely.
I mean, obviously nobody would ever do that *cough cough*.
"Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"
(Score: 2) by quacking duck on Tuesday August 12 2014, @02:09AM
I often jack-rabbit start, but look ahead, anticipate, and slow well in advance of a stop if I'm not blocking anyone behind me (e.g. they're trying to reach an advance green turn lane). I speed a bit, but never tailgate. I take curves a little harder when I'm driving by myself, but I've not been in an accident or spun out during winter driving (yet, knock on wood, etc).
Exactly what risk category would I be dump into?
(Score: 2) by tempest on Tuesday August 12 2014, @03:08PM
High risk because you're driving habits get lumped in with statistics, and they don't account for exceptions. Honestly I see no correlation with jackrabbit starts compared to the other hazards on the road. People who are texting or spaced out while on the phone do not typically jump on the accelerator, usually the opposite. For that matter I've noticed some of the people who drive the slowest, also frequently run stop signs and react to things which have nothing to do with them on the road (and often too late for that matter). As soon as I see "break tappers" I back way off.
(Score: 2) by darkfeline on Tuesday August 12 2014, @09:16PM
Jackrabbit starts are hard on your car (though I'm not a mechanic, so don't take my word for it), thus your car will need more maintenance and the components are more likely to be damaged, so you are at more risk of an accident and thus will be charged more for insurance.
Join the SDF Public Access UNIX System today!