Slashdot Mirror
Sensor Characteristics Uniquely Identify Individual Phones
An anonymous reader writes "SFGate reports that Stanford researchers have figured out a way to generate a unique fingerprint from a cell phone's suite of built-in sensors. The tiny accelerometers, gyroscopes, microphones, and speakers in cell phones have characteristics that vary slightly from handset to handset, and these variations may contain enough information to uniquely identify a given handset. How that information might get from the phone to a third party varies (the article describes a JavaScript snippet reading the Z-axis accelerometer, though it says little about how the user might block such information from being read), but the possibility for abuse is certainly troubling."
Now I have to drop my phone from time to time to fool the NSA.
Cell phones have been identifiable by RF fingerprinting for many, many years.
Was a common anti-fraud technique in the analog cellular days.
The possibility for abuse is troubling. Really?
Android: android.telephony.TelephonyManager.getDeviceId()
iOS: NSString* uniqueID = [UIDevice currentDevice].uniqueIdentifier;
WindPhone: Dunno don't do anything for it, I assume it's part of the API as well.
So yes, tell me more about this "troubling" ability to build a fingerprint of questionable accuracy on a device to uniquely ID it even when you can just READ THE UNIQUE DEVICE ID right from it to start with.
How long before we have Minority Report type crimes?
"Sir, you're going to have to have to come with us. Our metadata surveillance indicates you are likely to commit a crime, and our tracking of your phone indicates you were recently at a hardware store. We need to take you to the internment camp."
Some days I just want to turn into Reg the Blank and hide.
When they can know everything about you even when you've done nothing wrong, you're not so much free anymore as you are being allowed to pretend you are until such time as they decide to cart you off.
Lost at C:>. Found at C.
"Code running on the website in the device’s mobile browser"
So what I'd like to know is this (for all you people out there who write web code for mobile devices): what are the differences between what access different platforms give to those sensors? Obviously Android provides all the access that's needed; the example in the article refers to it working on a Galaxy Nexus. But what about Windows Mobile/IOS/Blackberry? Do they all have APIs to expose that functionality to something running in a browser, given that some of those platforms lack either java or shockwave/flash?
For your security, this post has been encrypted with ROT-13, twice.
"Uniquely identify a given handset"?
The MAC. QED.
A statistical analysis of your online writing-style identifies you. CCTV cameras identify you from your gait (the "way you walk"). And now your smartphone sensors give away what smartphone you are using (... useful to "backdoor" the device, I presume?). My question to these scientist: Why do you create this tech? Do you not care about the privacy of the common man, or indeed the technological future your children will be forced to live in? My 2 Cents on this, and similar efforts to "ID people"....
Why did the chicken cross the road? Because Elon Musk put an AI chip in its head.
It isn't sufficient anymore to block apps from getting some information (besides, some apps play miffed if they don't get this or that). What the OS should do is empower the user to tell "controlled lies".
For example: fuzz geo data by some controllable amount (or "I'm always in Trondheim, Norway"). Fuzz accelerometer, voice, and so on.
By default apply some sensible random fuzzing (just a tad above the instrument's accuracy, for example). Make the "lying strategies" configurable per-app.
I was of the impression that anything that accesses the cell network already has a unique IMEI adddress and that devices that access networks have a unique MAC address. What does this provide that they don't? It would seem this information could be spoofed at least as easily as such hardware addresses.
Every mobile phone, GSM modem or device with a built-in phone / modem has a unique 15 digit IMEI number.
Because there are lots of people who want PhD:s, but not a lot of creativity to go around and even less funding to go around for creative and truly novel projects
You can bet that this has already been done in the industry so it's not like they're inventing anything that doesn't already exist.
By the way, it ought to be reasonably straightforward to get a fingerprint out of the totality of sensor data that a phone generates during the course of a week or so even if the sensors were flawless. After all, we all have different habits, different gaits, etc. Odds are someone is already doing that.
I recall reading something earlier in the year that researchers (maybe it was our friends at NS*) are able to uniquely identify cell phones based on some type of timing difference in gsm trasmissions. Bottom line, if you use a phone someone can figure out its you.
If you look at the graph in the article (which talks about flipping the phone, but seems to actually be measurements of flat vs standing vertical), the variations are constrained to be (in the Sz axis) from 0.994 to 1.004, or a variation of 0.008, and the Sz repeatability is worse than 0.00025. So, this would work if the number of phones was ~ 30, but would be "confusion limited" for a larger number. Likewise, in the Oz axis the (different ?!?) units run from -0.2 to 0.4, a variation of 0.6, and the uncertainty is > 0.02, so the number of phones that could be distinguished is ~ 30. Combine these two axes, and no more than ~ 30^2 or 900 phones could be distinguished. There are obviously more than 900 phones in the world.
Even if all 3 sensors are independent and equally sensitive, that only gets you the ability to track 900^3 or ~ 700 million devices, which is a lot, but still likely not enough, as the distribution of errors is not likely to be uniform, but gaussian or some other distribution, and that will lower the effective sensitivity, as would any correlation between the sensor errors.
Note also that quartz crystals (I believe that these are piezoelectric sensors) are notorious not only for being individually imperfect, but also for drifting with time and (especially) temperature, which might also substantially reduce repeatability.
So, I suspect this is not likely to work well in practice.
What this could do is make the rare phone (one with by chance a particularly bad sensor) easily identifiable...
Pity their research was so slow. Steve Gibson of grc.com and the "Security Now!" podcast is in talks with the W3C about his new SQRL authentication protocol. Uniquely individual, completely anonymous.
"Give a man fire, and he'll be warm for a day; set a man on fire, and he'll be warm for the rest of his life
The tricky bit is remembering to change the speed setting every morning...
99% of us agree with you.
I come here for the love
IAMA scientist who creates such things. So here's my answer to your question: we create this kind of tech to allow law enforcement to identify individuals (in a very broad sense of all these terms), so we can lock them in (this is supposed to be very unsurprising).
If the tech in question is "fingerprint" (real ones, with your fingers), law enforcement is "police" (and not military/counter-terrorism/political) and individual is "criminal", I think pretty much everybody agrees that it is a good thing (you might be tempted to say otherwise, but imagine we're talking about someone you know/love having been assaulted/killed).
Crime happens where people are (e.g. homes, train stations, internet), and criminals use the same tools as we all do (e.g. screwdrivers, cutters, smartphones, etc.). If criminals move, law enforcement must be able to follow them (that's why police officers have powerful cars that exceed speed limits, btw). If the criminals start using smartphones, law enforcement starts using smartphones as a mean to identify/follow/[...] them. Or let them go - but this is something you'll have to explain to your children when they get robbed (or worse).
So, there is nothing new in creating new identification means - it has always happened, and will always do.
Now, the real concern is the way this kind of technology can be misused and abused (e.g. by governments or secret agencies). The question is not new at all, and people from all generations have had to take a stance on this - most of the times, in a democratic, free country, by going for a middle-ground approach (e.g. we collect the DNA of offenders, but are not allowed to keep them more than X years, and an independent supervisor makes sure the data does not leak, etc.)
The same reason that white hat security researchers look for holes in software. Sure, finding those holes and eventually releasing patches can help hackers identify exploits that might still be unpatched on some machines, but *not* finding those holes doesn’t mean they automatically go unfound. If a white hat didn’t find & announce it, there’s still a pretty good chance a black hat (or the NSA...) found it and is exploiting it in the wild. I’d honestly rather have a zero-day with a patch “coming soon” than have no idea there’s even a bug that’s being actively exploited without anyone knowing about it and no patch forthcoming.
Looking for this type of unique tracking capability in devices is valuable because it helps understand what the threat model of carrying one is. I’ve no doubt the NSA has a division looking for exactly this kind of “attack” against devices. If device & OS manufacturers care at all about privacy (debatable...), knowing this type of situation exists is the first step in attempting to mitigate it.
The fix is trivial, just add a small random X Y and Z bit of fuzz to the measurements, less than the sensor's natural variations, but enough to fuzz the results.
Also the sensors have a bit of natural temperature dependence, around 0.5 milliG's per degree C, so the readings are not all that good an identifier.
It's a simple equation
A) If you don't like tracking , just dont use it. Simple or ,making clear it guides your vote , the candidates about privacy laws and if they that have bills to put on the table.
B) No amount of whining will do anything to change that. Vote for someone who will introduce amendments to the privacy laws.
Use the power of your vote. Ask
As long as there are insufficient numbers of voters opposing the companies who fund the candidates , why would they do anything about it ?
A statistical analysis of your online writing-style identifies you. CCTV cameras identify you from your gait (the "way you walk"). And now your smartphone sensors give away what smartphone you are using (... useful to "backdoor" the device, I presume?). My question to these scientist: Why do you create this tech? Do you not care about the privacy of the common man, or indeed the technological future your children will be forced to live in? My 2 Cents on this, and similar efforts to "ID people"....
The very same ideas are used in other applications, for example authorship determination (eg did Shakespeare write Hamlet). This is how Science works. Progress in one area (eg identifying people from Smart Phones) will improve a host of other, related technologies, many of which are of humanitarian benefit.
The nice thing about a person's actual fingerprints is that they don't change over time. As one poster pointed out, oscillators do drift over time. I can't help but think that the components they're trying to measure also will change in the tested characteristics as they age. If a digital fingerprint doesn't stay constant over the life of the device, is it really of any value?
By comparing the frequencies of the various crystals in the PC. Processor vs. serial port, etc. An attempt at keeping over seas customers from ghosting one paid installation onto hard drives in a massive rollout.
Before that, when alternate track tables existed on hard drives, that was another form of identification. Then there were less useful ones like RAM size, video adapter type, list of boards in the system, etc.
That was not exact enough to give anything like certainty. The quick correlation of CPU to serial crystal for instance, only gave about 2-3 bits of identity.
Now if you have an image sensor, and you can see the dark values of the pixels, that would be a real fingerprint.
Accels and gyros, too noisy, and must be motionless to get anything useful. OK if sitting on a table, but not inside a car.
Businesses want to track you because there's money in data mining and profiling. Governments want to track you for surveillance and control. You think you'd be one iota less tracked if nobody in academia did? No, you'd just not realize it but I guess ignorance is bliss...
Live today, because you never know what tomorrow brings
"Because if I don't do it, somebody else will"
Interesting how this provides potential for both security and privacy invasion.
Because it's an easy publication and most people, scientists or not, are unethical.
Yeah, let's blame it on the scientists who publish it. Like we scapegoat the whitehats that report vulnerabilities in software. I cannot tell if you're kidding or not. Hopefully you can tell, now, that reporting these results means that people need to sanitize this information, and/or demand that manufacturers help sanitize/restrict access to this data.
If you didn't know it, the "bad guys" would still know it, and be using it without your knowledge.
The sonar guy went on to become an ADA in New York.
If you need to go and murder someone, leave your phone at home.
iirc, same as a technique proposed about 6 years ago to ID computers ??