Smartphone sensors leave trackable fingerprints

Example demonstrating how accelerometer data shared with separate traffic and health applications could indicate Bob's location. Credit: Image courtesy of University of Illinois College of Engineering

Fingerprints -- those swirling residues left on keyboards and doorknobs -- are mostly invisible. They can affirm your onetime presence, but they cannot be used to track your day-to-day activities.

They cannot tell someone in real time that after exercising at the gym, you went to office in a bus and played video games during lunch. But what if our hand-held electronics are leaving real-time fingerprints instead? Fingerprints that are so intrinsic to the device that, like our own, they cannot be removed?

Research by Associate Professor Romit Roy Choudhury and graduate students Sanorita Dey and Nirupam Roy has demonstrated that these fingerprints exist within smartphone sensors, mainly because of imperfections during the hardware manufacturing process.

In some ways, it's like cutting out sugar cookies. Even using the same dinosaur-shaped cutter, each cookie will come out slightly different: a blemish here, a pock there. For smartphone sensors, these imperfections simply occur at the micro- or nanoscale.

Their findings were published at the Network and Distributed System Security Symposium (NDSS), a major conference on wireless and web security, held last February in San Diego. 

The research also won the best poster award at the HotMobile international workshop in 

2013.

The researchers focused specifically on the accelerometer, a sensor that tracks three-dimensional movements of the phone -- essential for countless applications, including pedometers, sleep monitoring, mobile gaming -- but their findings suggest that other sensors could leave equally unique fingerprints.

"When you manufacture the hardware, the factory cannot produce the identical thing in millions," Roy said. "So these imperfections create fingerprints."

Of course, these fingerprints are only visible when accelerometer data signals are analyzed in detail. Most applications do not require this level of analysis, yet the data shared with all applications -- your favorite game, your pedometer -- bear the mark. Should someone want to perform this analysis, they could do so.

The researchers tested more than 100 devices over the course of nine months: 80 standalone accelerometer chips used in popular smartphones, 25 Android phones, and 2 tablets.

The accelerometers in all permutations were selected from different manufacturers, to ensure that the fingerprints weren't simply defects resulting from a particular production line.

With 96 percent accuracy, the researchers could discriminate one sensor from another.

"We do not need to know any other information about the phone -- no phone number or SIM card number," Dey said. "Just by looking at the data, we can tell you which device it's coming from. It's almost like another identifier."

In the real world, this suggests that even when a smartphone application doesn't have access to location information (by asking "this application would like to use your current location"), there are other means of identifying the user's activities. It could be obtained with an innocuous-seeming game or chatting service, simply by recording and sending accelerometer data. There are no regulations mandating consent.

To collect the data, the researchers -- as with any would-be attacker -- needed to sample the accelerometer data. Each accelerometer was vibrated using a single vibrator motor -- like those that buzz when a text message is received -- for two-second intervals. During those periods, the accelerometer detected the movement and the readings were transmitted to a supervised-learning tool, which decoded the fingerprint.

"Even if you erase the app in the phone, or even erase and reinstall all software," Roy said, "the fingerprint still stays inherent. That's a serious threat."

At this point, however, there is no absolute solution. Smartphone cases made of rubber or plastic do little to mask the signal. Deliberately injecting white noise in the sensor data can smudge the fingerprint, but such noise can also affect the operation of the application, making your pedometer inaccurate and functionally useless.

If accelerometer data were processed directly on the phone or tablet, rather than on the cloud, the fingerprint could be scrubbed before sending information to the application.

That is, the pedometer application might only receive basic information like "300 steps taken," rather than receiving the raw accelerometer data. This, however, imposes a load on the phone's processor and, more importantly, reduces the phone's battery life.

The research also suggests that other sensors in the phone -- gyroscopes, magnetometers, microphones, cameras, and so forth -- could possess the same types of idiosyncratic differences. So even if, at a large scale, the accuracy of accelerometer fingerprints diminishes, when combined with prints from other sensors, an attack could be even more precise.

"Imagine that your right hand fingerprint, by some chance, matches with mine," Roy Choudhury said. "But your left-hand fingerprint also matching with mine is extremely unlikely. So even if accelerometers don't have unique fingerprints across millions of devices, we believe that by combining with other sensors such as the gyroscope, it might still be possible to track a particular device over time and space."

For smartphone users and e-book readers, smartwatch wearers and tablet devotees, perhaps the most critical take-home message, in the short run anyway, is the importance of vigilance.

"Don't share your accelerometer data without thinking about how legitimate or how secure that application is," Dey said. "Even if it's using only the sensor data, still it can attack you in some way. The consumer should be aware."

Source: University of Illinois College of Engineering

Cats and humans have shared the same households for at least 9,000 years, but we still know very little about how our feline friends became domesticated.

The scientists fitted GPS collars and motion sensors on 38 free-ranging lynx for the study. Credit: Image courtesy of Albert-Ludwigs-Universität Freiburg

An international research team recorded and analyzed the activity patterns of 38 wild cats over the course of months Whether a lynx hunts by day or by night and how active it is overall depend primarily on the behavior of the wild cat's most important prey and its individual traits -- lighting conditions, on the other hand, do not play a major role in its basic behavioral patterns. This is the key finding of a study published in the journal PLOS ONE by an international research team led by forest scientist Dr. Marco Heurich.

The scientists fitted GPS collars and motion sensors on 38 free-ranging lynx for the study. Since the study sites were located across a wide latitudinal range from Central Europe to northern Scandinavia, the length of days and nights varied greatly between them. The team recorded and analyzed the activity patterns of the wild cats on a total of more than 11,000 days. The results reveal that lynx in more southerly regions are most active at dawn and dusk and that they move more by night than by day. They take their longest break in the middle of the day, and this break is extended as daylight duration increases. However, the cats exhibit this basic behavioral pattern independently of lighting conditions: "Lynx keep to a 24-hour rhythm with an active and a resting phase even on the polar day and the polar night," reports Heurich.

What the study found to be more important for explaining the wild cats' activity patterns are their individual traits: Young lynx are more active than adult lynx, and male adults are more active than female adults. In addition, they move more in spring and summer than in fall and winter, and the farther north they live, the larger the territory they cover -- and this of course results in higher activity. Lynx adapt their hunting schedule to the behavior of their prey. In polar regions, the height of their activity at dusk is less pronounced. 

This corresponds to the behavioral pattern of reindeer, which exhibit a steady movement profile outside of their sleeping phases.. In Central Europe, by contrast, the team found a maximum amount of activity at dusk -- in lynx as well as in deer. "The findings of this study make an important contribution to our understanding of the habits of predatory animals in our landscape," says Heurich. "They also show that human activities in the areas included in the study do not have a general influence on the activity pattern of the animals."

Subtle shifts in the Earth could forecast earthquakes, tsunamis

University of South Florida graduate student Denis Voytenko prepares a GPS unit for a high-precision geodetic measurement.
Credit: Jacob Richardson

Earthquakes and tsunamis can be giant disasters no one sees coming, but now an international team of scientists led by a University of South Florida professor has found that subtle shifts in Earth's offshore plates can be a harbinger of the size of the disaster.

In a new paper published in the Proceedings of the National Academy of Sciences, USF geologist Tim Dixon and the team report that a geological phenomenon called "slow slip events" identified just 15 years ago is a useful tool in identifying the precursors to major earthquakes and the resulting tsunamis. The scientists used high precision GPS to measure the slight shifts on a fault line in Costa Rica, and say better monitoring of these small events can lead to better understanding of maximum earthquake size and tsunami risk.

"Giant earthquakes and tsunamis in the last decade -- Sumatra in 2004 and Japan in 2011 -- are a reminder that our ability to forecast these destructive events is painfully weak," Dixon said.

Dixon was involved in the development of high precision GPS for geophysical applications, and has been making GPS measurements in Costa Rica since 1988, in collaboration with scientists at Observatorio Vulcanológico y Sismológico de Costa Rica, the University of California-Santa Cruz, and Georgia Tech. The project is funded by the National Science Foundation.

Slow slip events have some similarities to earthquakes (caused by motion on faults) but release their energy slowly, over weeks or months, and cannot be felt or even recorded by conventional seismographs, Dixon said. Their discovery in 2001 by Canadian scientist Herb Dragert at the Pacific Geoscience Center had to await the development of high precision GPS, which is capable of measuring subtle movements of the Earth.

The scientists studied the Sept. 5, 2012 earthquake on the Costa Rica subduction plate boundary, as well as motions of the Earth in the previous decade. High precision GPS recorded numerous slow slip events in the decade leading up to the 2012 earthquake. The scientists made their measurements from a peninsula overlying the shallow portion of a megathrust fault in northwest Costa Rica.

The 7.6-magnitude quake was one of the strongest earthquakes ever to hit the Central American nation and unleased more than 1,600 aftershocks. Marino Protti, one of the authors of the paper and a resident of Costa Rica, has spent more than two decades warning local populations of the likelihood of a major earthquake in their area and recommending enhanced building codes.

A tsunami warning was issued after the quake, but only a small tsunami occurred. The group's finding shed some light on why: slow slip events in the offshore region in the decade leading up to the earthquake may have released much of the stress and strain that would normally occur on the offshore fault.

While the group's findings suggest that slow slip events have limited value in knowing exactly when an earthquake and tsunami will strike, they suggest that these events provide critical hazard assessment information by delineating rupture area and the magnitude and tsunami potential of future earthquakes.

The scientists recommend monitoring slow slip events in order to provide accurate forecasts of earthquake magnitude and tsunami potential.

Source: University of South Florida (USF Health)

The Severe drought is causing the western US to rise like a spring uncoiling

Maps of GPS points in the western United States, with blue indicating a drop and yellow-red reflecting a rise.
Credit: Image courtesy of University of California - San Diego

The severe drought gripping the western United States in recent years is changing the landscape well beyond localized effects of water restrictions and browning lawns. Scientists at Scripps Institution of Oceanography at UC San Diego have now discovered that the growing, broad-scale loss of water is causing the entire western U.S. to rise up like an uncoiled spring.

Investigating ground positioning data from GPS stations throughout the west, Scripps researchers Adrian Borsa, Duncan Agnew, and Dan Cayan found that the water shortage is causing an "uplift" effect up to 15 millimeters (more than half an inch) in California's mountains and on average four millimeters (0.15 of an inch) across the west. From the GPS data, they estimate the water deficit at nearly 240 gigatons (62 trillion gallons of water), equivalent to a six-inch layer of water spread out over the entire western U.S.

Adrian Borsa, an assistant research geophysicist at Scripps Institution of Oceanography, UC San Diego.

Results of the study, which was supported by the U.S. Geological Survey (USGS), appear in the August 21 online edition of the journal Science.

While poring through various sets of data of ground positions from highly precise GPS stations within the National Science Foundation's Plate Boundary Observatory and other networks, Borsa, a Scripps assistant research geophysicist, kept noticing the same pattern over the 2003-2014 period: All of the stations moved upwards in the most recent years, coinciding with the timing of the current drought.
Agnew, a Scripps Oceanography geophysics professor who specializes in studying earthquakes and their impact on shaping Earth's crust, says the GPS data can only be explained by rapid uplift of the tectonic plate upon which the western U.S. rests (Agnew cautions that the uplift has virtually no effect on the San Andreas fault and therefore does not increase the risk of earthquakes).

For Cayan, a research meteorologist with Scripps and USGS, the results paint a new picture of the dire hydrological state of the west.

"These results quantify the amount of water mass lost in the past few years," said Cayan. "It also represents a powerful new way to track water resources over a very large landscape. We can home in on the Sierra Nevada mountains and critical California snowpack. These results demonstrate that this technique can be used to study changes in fresh water stocks in other regions around the world, if they have a network of GPS sensors."

The study was supported by USGS National Earthquake Hazards Reduction Program

Source: University of California - San Diego