Turns Out Cell Phone Location Data Is Not Even Close To Accurate, But Everyone Falls For It
from the because-of-course dept
We've frequently talked about law enforcement and the intelligence community accessing and making use of cell site location data, which looks to figure out where people are based on what cell towers they're connected to. Law enforcement likes to claim that it doesn't need a warrant for such data, while the NSA has tested a pilot program recording all such data, and says it has the legal authority to collect it, even if it's not currently doing so.However, as anyone with even a basic geometry education recognizes, which cell tower you're connected to does not give you a particularly exact location. It can be useful in putting someone in a specific (wide) area -- or, much more useful in detailing where someone is traveling over long distances as they repeatedly switch towers in a particular direction. But a single reading does not give you particularly exact location details. I had naturally assumed that most people understood this -- including law enforcement, lawyers, prosecutors and judges -- but it turns out they do not. A rather depressing story in The Economist notes that, thanks to this kind of ignorance (combined with bogus cop shows on TV that pretend cell site data is good for pinpointing locations), cell site location data is frequently used to convict innocent people. The story opens with a ridiculous example, in which a woman was pressured into a plea bargain based on totally false claims about tower location data:
SOMEONE strangled a prostitute in Portland, Oregon in 2002. The police arrested Lisa Roberts, the victim’s ex-lover, who spent more than two years in custody awaiting trial. Shortly before the trial the prosecutor told Ms Roberts, via her lawyer, that tower data collected by Verizon, her mobile-telephone network, showed precisely where she was at the time of the murder. As her lawyer recalled, the prosecutor said Ms Roberts could be “pinpointed” in a park shortly before the victim’s naked and sexually assaulted corpse was found there. She was told she faced 25 years to life in prison. She accepted a deal to plead guilty and serve 15 years.Obviously, things like GPS do allow for much more precise targeting of location (which may be why the NSA is focusing on that instead of cell site location data), but too many people confuse cell site location data with GPS. What's ridiculous is that this mistake isn't just being made by random people -- but prosecutors and lawyers responsible for criminal cases that can destroy an innocent person's life.
But the high-tech evidence against her was bunk. Routinely collected tower data can place a mobile phone in a broad area, but it cannot “pinpoint” it. That would require a special three-tower “triangulation”, which cannot reveal past locations. It took a decade for Ms Roberts’s guilty plea to be thrown out. On May 28th she left prison, her criminal record clean, after nearly 12 years in custody.
This really points to a larger issue: people have this tendency to believe that technology can answer all questions. The NSA's fetishism of surveillance via technology is an example of this. There's data there, so it becomes all too tempting to assume that the data must answer any possible question (thus, the desire to collect so much of it). But the data and the interpretations it can lead to are often misleading or simply wrong. And that's especially true when dealing with newer technologies or forms of data collection. That the criminal justice system could go decades without everyone recognizing the basic geometric limits of cell site location data based on a single cell is... both astounding and depressing. But it's also a reminder that we shouldn't assume that just because some evidence comes from some new-fangled data source it's automatically legitimate and accurate.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: accuracy, cell site location data, gps, location data, plea bargains, prosecutions
Reader Comments
Subscribe: RSS
View by: Time | Thread
Anyway back to the story, it's good to rise awareness of these issues because people will be better equipped to counter such "evidence". Also, this pleading guilty thing seriously needs to change in the US.
[ link to this | view in chronology ]
Re:
The short answer is how many GPS signals you're getting.
"In typical GPS operation, four or more satellites must be visible to obtain an accurate result."
http://en.wikipedia.org/wiki/Global_Positioning_System
GPS can be just as bad at pinpointing an exact location as cell site location if you don't have 4 or 5 satellites in view. With 4 clear signals, accuracy is 6-12 meters. Can't find specifics on less, but I know I've been in situations where my phone thinks I'm half a mile or more away from where I actually am (frequently indoors in a very large building, so probably getting less than the 4 signals).
[ link to this | view in chronology ]
Re: Re: GPS Accuracy
[ link to this | view in chronology ]
Re: Re: Re: GPS Accuracy
[ link to this | view in chronology ]
Re: Re: Re: Re: GPS Accuracy
If you're at a fixed known location you can calculate what signal you should be receiving from any one GPS signal. It in essence can confirm your location, but as you already knew where you were there isn't a lot of value in it. Unless you are checking to see if you've moved. In which case you'll confirm you know where you are, or you'll know you're lost.
[ link to this | view in chronology ]
Re: Re: Re: Re: GPS Accuracy
[ link to this | view in chronology ]
Re: Re: Re: GPS Accuracy
GPS uses times being sent by satellites that are in known positions. The clock in your phone is nowhere near accurate enough to use, so the times sent from the satellites are actually used to determine what time it is at your location as well as the distance to the satellite. GPS receivers assume you are on the face of the planet, so if you knew the time, you would know the distance to two satellites and your location, but you need at least a third to determine the time and then would need a fourth to stop assuming the altitude. After that, additional satellites will help your device calculate a more accurate time at your location and that will make the distance to each satellite more accurate.
As for the cellular end of this - the cell towers do not transmit hyper-accurate time information and their locations to your handset, so at best you can get intersection information based on signal strength - the time the signal takes to get to the cell phone cannot be measured by the equipment in the phone accurately enough to be worth anything.
What IS happening now is the phones are using GPS technology for E911 and transmitting it back to the towers so the towers do actually have an accurate picture of where the phone is.
[ link to this | view in chronology ]
Re: Re: Re: Re: GPS Accuracy
Have you ever heard of "timing advance"? Yes, it measures the time for the signal to get to the cell phone. It is used as a timing adjustment (thus the name) so that, when the signal arrives on the tower, it does not step on the signal sent by other phones nearer/farther the tower.
It's easy to measure the distance with radio: send a "ping" to the other side, wait for the answer, and divide the time it took by two. That method is even used to locate deep space probes with high precision.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: GPS Accuracy
Assuming the signal travels at 300,000,000 meters per second (it doesn't since it's not in a vacuum but this is ballpark), to get 10 meter resolution you need a clock accurate to 1/30,000,000 of a second. Are the clocks in cell phones that precise?
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: GPS Accuracy
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re:
As for the phone I think 6-12 meters seem reasonable considering it usually can't pinpoint which lane you are when you have two parallel roads.
[ link to this | view in chronology ]
Re: Re: Re:
Yes, this is possible.
The way GPS works is that all the GPS satellites send out a synchronized timecode (generated by an on-board atomic clock) that has a precision of around 100 nanoseconds, if I remember correctly. Since all the satellites are sending the same timecode at the same moment, the receiver simply identifies when it has received the timecodes from all the satellites it can see, then compares the actual time they were received with the time they were sent (the timecode itself). The discrepancies between the timecode and when the signal was received are then compared and used to "triangulate" (it's not really triangulation, but similar to that). US GPS satellites used to intentionally introduce a random fluctuation to artificially limit the accuracy, but they stopped doing that a number of years ago.
So, the quality of the receiver affects accuracy in a number of ways, from how many satellite signals it can receive, to how quickly and accurately it can do the math, to how accurate its timekeeping is.
Inexpensive GPS receivers (such as in phones) are not terribly accurate.
[ link to this | view in chronology ]
Re: Re: Re: Re:
Actually, I was wrong. The actual limit, if you're only using GPS signals, is 10cm. However, you can increase this resolution by using things like fixed ground-based GPS refrerences.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Cell phone location
[ link to this | view in chronology ]
Source: I have worked with E911 services.
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re:
Source: I have worked with E911 services.
Could you expand on that a bit? My understanding is that when there's multiple towers, the signals can be triangulated and the location of the phone can be calculated accurately. That's how, for example, the map application in a phone knows where the phone is located. Mike says "too many people confuse cell site location data with GPS," but I don't believe phones even have GPS. Is that right?
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
The chip to get GPS signal is really tiny, from there it's just processing. I had a USB one for my computer that was smaller than my watch and it was mostly plastic for extra protection. My laptop at the time was less powerful than my watch.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Re:
There are apps for Android that do download entire map packages, but last I saw they were the on the old business model. You had to pay for the map and then pay for each update.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re:
Most allow you to do this pretty easily. Google maps is one of the exceptions to this -- but there are numerous alternatives.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re:
http://en.wikipedia.org/wiki/Assisted_GPS
agps data is passed to e911 so i would assume it was what police use when they get a warrent in most cases. Im sure there are time out in the country where the phone cant sync the timeing signals from the three closest towers but for most metropolitan areas I would say trigulation by cell phone is very accurate.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
You can say that, but there is no evidence that this is true and a lot of evidence that it is not. Even the law doesn't require anything like great accuracy.
[ link to this | view in chronology ]
AGPS versus GPS
AGPS is full blown satellite GPS, but with location and (especially) startup assisted by cell tower location information, as full blown GPS startup can take over 30 seconds when the GPS has no idea where on earth you may be.
[ link to this | view in chronology ]
Re: Re:
By the way, those requirements are: carriers using 'handset based' (GPS) technology must report handset location within 50 meters for 67% of calls, and within 150 meters for 90% of calls; carriers using 'network based' (tower triangulation) technology must report location within 100 meters for 67% of calls and 300 meters for 90% of calls
[ link to this | view in chronology ]
Re: Re:
People however do confuse GPS and Tower triangulation - they are two separate things, both of which can locate where a device is being used.
As far as the triangulation goes, if someone is in a location where they can reach multiple towers (both of their own carrier or others) you can get it down to about a 6ft diameter area of where a person is using the tower data alone.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Article in accuracies
The triangulation previously done to place analog and pre-smart phones was truly not so accurate but if you have a smartphone, either android or iphone then your phone CONTINOUSLY gives updates to the closest tower your phone location, handset signal strength and statistics on any dropped calls. These are evaluated daily by carriers to determine coverage and capacity issues.
Federal Legislation under enhanced E-911 requires an accuracy within 100 ft of the handset.
[ link to this | view in chronology ]
Re: Article in accuracies
There is a difference between a live query, where is this phone now, and the querying the much more limited data that the phone company stores. I would expect, and the statement suggests, that only very limited 'event data' is actually stored, like time and tower when a phone connects to and disconnects from a tower.
This makes perfect sense because of the orders of magnitude, 3 or 4, between 'event data' and second by second data from several towers. This may if useful to the company, include location data for the event, when they wish to survey coverage, or locate the boundaries of a black spot in there coverage. Otherwise they would be storing data that they do not need, and that eats into their profits. This would not preclude them using live location data, if they could use it to sell customers to advertisers, but they would have no reason to store the data.
[ link to this | view in chronology ]
Re: Re: Article in accuracies
This varies from company to company (along with the retention period of records) but normally it is a single tower location, a time and an antenna ID. Most towers are composed of three antennas offset to provide 360 degrees of coverage, so each antenna ID corresponds to about 120 degrees of sweep. Analysts can evaluate tower location for elevation, obstructions, and competing towers to provide rough estimates for the maximum distance from which a device is likely to connect. It is generally several miles.
This provides a large swath of coverage that the phone is deduced to have been inside of at that time. If the device quickly "bounces" back and forth between that and another tower (or two) the coverages may be overlaid to perform a crude "triangulation" but nothing close to GPS. The odds that a device would bounce during any specific event, and that all the records would be kept, are pretty low.
This information comes from my having served on a federal jury and being presented with phone record location information and analysis. I was very surprised at the time because I assumed it would be much, much better.
[ link to this | view in chronology ]
Mathematics
As for the signal travel time - I highly doubt the dollar store towers would have that capability.
You are stuck with needing a minimum of two to triangulate as long as you can get a direction.
As for GPS. That is kyboshed to a 10 meter error by the USA. You can get better with the right surveying software and time OR receive the Russian signals OR wait for the EU to finish their system.
There is also ground based nodes in some cities that compensate for blocking LOS to the satellites.
[ link to this | view in chronology ]
Re: Mathematics
This hasn't been true for a few years now. Now, all of the non-receiver error is due to noise and 95% of all errors are between 0 and 10 meters. Even that error can be reduced or eliminated by using fixed groundpoints.
[ link to this | view in chronology ]
Plea bargain system
Yes, false information. Don't think law enforcement and prosecutors didn't know they were spieling nonsense. They knew exactly how inaccurate the information was; they just didn't care, as long as it yielded a confession.
[ link to this | view in chronology ]
Re: Plea bargain system
[ link to this | view in chronology ]
Re: Plea bargain system
[ link to this | view in chronology ]
Re: Plea bargain system
Agreed, pleaing some guilt rather than risk a false conviction by over-zealous prosecutors is not what our justice system is supposed to do. Unfortunately, I feel that I have too much cynicism in the authorities to ever be allowed on a jury.
[ link to this | view in chronology ]
Re: Re: Plea bargain system
It's impossible to say without actually being in the situation, of course, but this wouldn't be such a slam-dunk decision for me.
First, 15 years is a substantial amount of time. Aside from that, though, a murder conviction effectively destroys your life whether you spend the next 15 years or the next 100 years in prison. You're pretty much done for either way.
[ link to this | view in chronology ]
Re: Plea bargain system
[ link to this | view in chronology ]
GPS on phones
However, the phone I had in 2002 had no GPS capability whatsoever.
[ link to this | view in chronology ]
just use multiple towers?
Can more than one cell tower "see" a phone and its signal strength at a given time? Or does the phone simply look for cell signals passively to decide which tower to connect to, meaning that only the connected tower sees the phone? If more than one cell tower can "see" a phone at a given time, the phone company would then be able to nail my phone's location to the same precision as I see on my phone, right?
[ link to this | view in chronology ]
Re: just use multiple towers?
Yes, all towers that are in range of your phone has this information. In fact, they have to in order to make decisions about when and where to do handoffs to neighboring towers when the phone is in motion.
"If more than one cell tower can "see" a phone at a given time, the phone company would then be able to nail my phone's location to the same precision as I see on my phone, right?"
Technically, they could do so with greater precision than you see on your phone.
[ link to this | view in chronology ]
Re: Re: just use multiple towers?
Yes, all towers that are in range of your phone has this information. In fact, they have to in order to make decisions about when and where to do handoffs to neighboring towers when the phone is in motion."
Whenever a Techdirt article references cellular networks, the comments go to shit, as the clueless highlight their cluelessness. John Fenderson's comment (to pick just one) is a perfect example. *Phones* monitor signals from multiple sites and *the phones* decide when it's time to switch to another site. This is based on the received signal strength (RSSI) and the bit error rate (BER), factors which only *the phone* knows. FFS people...
" "If more than one cell tower can "see" a phone at a given time, the phone company would then be able to nail my phone's location to the same precision as I see on my phone, right?"
Technically, they could do so with greater precision than you see on your phone."
More than one cell tower can not see a phone at the same time, because (for most technologies) each tower is using its own TX and RX frequencies. What one tower can see, the others cannot. A system operator would have to target a specific phone and force it to talk to several sites in turn. Each site can time a round trip, which the backend can then use to triangulate.
As far as knowing where everyone and their phone has been, cellular systems just store what sites every phone registers to and when. This does not provide a very detailed picture, as phones only register when they transition from one service area to another, and service areas are comprised of dozens of neighboring sites, so in the city, a phone might travel 5+ miles before reregistering, and much greater distances when outside cities. Once a phone has registered, the system does not know, or care, where in the service area the phone is. Phones only register on service area changes because that is what the system uses to know which sites (the sites comprising your current service area) have to page your phone when you get an incoming call or message. So no, they're not logging everyone's location with any geographical precision or on a frequent schedule.
[ link to this | view in chronology ]
Cell location vs TDOA vs GPS for nontelecom types
There are three separate processes used to locate a cell phone: 1) cell (tower) the phone is connected to (I think this is what Masnick is referring to), 2) Time Difference of Arrival (TDOA) "triangulation," and GPS.
[Many people think of "triangulation" as directional antennas being used to take bearings on a source, based on signal strength, as in seen in movies. TDOA doesn't do this because cell towers do not have moveable directional antennas (you're reasonably safe in thinking of the array as an omnidirectional antenna). TDOA simply measures the time a cell phone signal takes to reach multiple towers. All the towers have their own GPS-based synchronized timing, so they can "confer" and agree that the signal each saw was the same connection attempt from the same phone, and compute the TDOA. Time gives distance. Distance gives a circle around each tower. The phone is where the arcs intersect.]
Number 1 is the easiest to derive (if the phone connects to the network, they perforce know what tower it's coming from), but least accurate; the phone could be anywhere in the circle area where they can get a signal.
Number 2 is reasonably accurate to several tens of meters, which is good enough for E911 work.
3 is quite accurate. If you have enough GPS birds in view. And you haven't had a glitch that puts you 100 kilometers off the Florida coast as you head into Chi-town.
I think most modern smart phones incorporate GPS (some years back, a bright company finally built complete GPS receivers on a single chip), but not all phones have it, and GPS doesn't always yield a good location. Cell companies adopted TDOA location both as an interim measure before GPS-capable phones became common and as a backup to GPS.
[ link to this | view in chronology ]
Interesting
[ link to this | view in chronology ]
Re: Interesting
[ link to this | view in chronology ]
So does that mean that my 1TB hard drive is actually storing a perfect copy of the 20TB+ I've downloaded over the last few years? Why can't I access all that data?
[ link to this | view in chronology ]
Re:
You have 1TB of data, the data you remove is flagged as available space but it is not actually removed until it is in fact overwritten with other data. Which is why you scrub mechanical drives with random data and even then you can recover some data if you're unlucky.
That's also why the NSA and other agencies have a mechanical destruction policy on all HDD's leaving facilities after being declared unservicable.
So all in all, i think the people doing that aren't that clueless about tech, you are.
So to answer your questions, No it isn't storing 20TB+, it's storing whatever is flagged as writable space.
You fill her up, remove all data (not scrub) then place 20GB on it, the remaining 980 GB still holds your previous data. Which is easily recovered by a few simple steps.
Better way to do it is : encrypt everything on the hard drive with AES, then scrub it. and if you're really paranoid, encrypt the drive again and scrub it again.
[ link to this | view in chronology ]
Vodafone Australia
[ link to this | view in chronology ]
I want you to remember this next time you're tempted to defend techno-utopianism.
[ link to this | view in chronology ]
My phone location ⇏ my location
[ link to this | view in chronology ]
single cell tower data vs. gps
[ link to this | view in chronology ]
guilty
[ link to this | view in chronology ]
Re: guilty
Because innocent people get convicted, and long prison sentences are scary.
Sonar does not use triangulation and is pretty precise.
True, but irrelevant, since cell towers and sonar detectors don't work the same way.
[ link to this | view in chronology ]
Why are some phones' GPS more accurate than others
We care because we have an Autistic son, and it's really important for us to be able to locate him sometimes. I need to replace his phone, and I'd like one with accurate GPS.
[ link to this | view in chronology ]
location
[ link to this | view in chronology ]
What is wrong
[ link to this | view in chronology ]
mistaken location
[ link to this | view in chronology ]
Re: mistaken location
[ link to this | view in chronology ]
I am right, location lies
My boyfriend will ask me about a certain address and street and I am appalled by the fact he believes these stupid phone apps. Huge fights over it. I'don't like that I'may unable to prove myself to be correct. Piss on these fight starting piece of shit apps.
But, it states location so it must be right. Bulshit.
[ link to this | view in chronology ]
Re: I am right, location lies
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Finding location of a cell phone in a broad area
[ link to this | view in chronology ]
Re: Finding location of a cell phone in a broad area
If you're not the owner, it's unlikely you'll be able to get this information.
[ link to this | view in chronology ]
Re: Re: Finding location of a cell phone in a broad area
[ link to this | view in chronology ]
Re: Phone was accurately tracked
That's why it could be tracked so accurately. This article is about using cell site data to locate a phone, which is less precise and less reliable.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Re:
It's neither, it's just that it doesn't always work that well.
[ link to this | view in chronology ]
GPS Accuracy
If you want Pin point accuracy you need 2 GPS's. On a Geographical Surveyed point. That feeds differential data back to the roaming GPS to the centimeter accuracy.
Did this for DOE Hanford. It was not use against Iraq. I quit when they put it in an Airplane.
[ link to this | view in chronology ]
The Author just outed his ignorance - probably Art-History Phd
To identify the location using towers, you do not use just one tower, you use, 2-4 towers. Any kids that passed high school electronics could tell you this. Based on signal strength changes, you can determine location and direction. This technology has been around since the 90's and used as far back as WWI to find radio towers, and illegal transmitters. It is NOT GPS, but can get you within a 1000 feet, when towers are close enough, like in an urban setting.
[ link to this | view in chronology ]
Re: The Author just outed his ignorance - probably Art-History P
If all you have is one tower, what you can tell about the location is that it's close enough to that tower to connect to it. So roughly a circular area, and a pretty large one. Here is what the author said:
"which cell tower you're connected to does not give you a particularly exact location. It can be useful in putting someone in a specific (wide) area"
Is there something about that statement that is incorrect?
I don't remember what Masnick's undergraduate degree is but he has an MBA, not an art history PhD.
[ link to this | view in chronology ]
Can track without gps
Gps operates on much simpler but similar technology. A log of a handoff between two towers and the time difference being corrected a such high carrier or even input rates could be used especially iff your phone is being physically examined for logs and software linked to these logs along with tower logs.
[ link to this | view in chronology ]