News

Overclocked Wearables Can Pick Up Bio-Acoustic Signals

November 25, 2016 by Dr. Steve Arar

Software updates can make conventional wearables capable of recognizing the wearer's movements.

Researchers have demonstrated that overclocking can make conventional wearables capable of recognizing the wearer's movements.

The sensors incorporated into wearables can sometimes be repurposed to perform tasks beyond their intended applications. For example, it's been shown that it's possible to discover a victim user’s passwords and PINs by applying a sophisticated algorithm to the data gathered by wearable embedded sensors.

Recently, researchers at the Future Interfaces Group at Carnegie Mellon University have overclocked the accelerometer of an LG smartwatch to extend its capabilities to more than just tracking fitness. By overclocking the off-the-shelf smartwatch via some software updates, they can now detect and process very small vibrations and audio signals.

The new technology, dubbed ViBand, can allow different apps to understand the context of your activities by capturing bio-acoustic signals.

The Applications of ViBand

The hacked watch can distinguish between different tools, such as a guitar, a drill, and more, as you use them. It can also interpret different gestures such as flicks, snaps, scratches, taps, and other motions. It even goes beyond this: ViBand can differentiate tapping on the forearm, the palm of the hand, or the back of the hand.

All of these are possible because each activity generates a distinctive bio-acoustic signal. As a result, via overclocking, you can use common gestures to control your smartwatch and all the objects which are connected to your watch through the IoT.

Tapping on different parts of the hand generates a bio-acoustic signal which is distinguishable to the ViBand. Image courtesy of CMU.

Gierad Laput, a Ph.D. student in the Human-Computer Interaction Institute (HCII), notes that the new technology allows your hand, which is a natural way of interacting with the world, to act as a detection device.

The research team proposes that several other interesting applications are possible, such as building object-aware apps which monitor your activities. For example, the device could recognize when you're preparing a meal and give advice if necessary.

Moreover, they propose a device called vibro-tag which utilizes inaudible vibrations to transmit data. Installed on an office door, a vibro-tag can transmit information—phone numbers, office hours, and more—to your watch.

Overclocking a Commercial Watch

Generally, in a wearable, the software restricts the sample rate of the embedded accelerometers to about 100Hz. This sample rate is sufficiently high for the common applications of today’s wearables such as activating the screen when the user lifts an arm or counting the footsteps.

However, to capture the bio-acoustic signals, a sample rate of 4kHz was required. The research team hacked the Linux kernel on the watch and upgraded the software. To examine their idea, researchers developed some experimental apps for ViBand. The new technology could successfully control the apps on the watch or the remote devices, such as the lights and a TV, through common gestures.

Drawbacks and Concerns

The main drawback to the hacked watch is a considerable increase in the power consumption. Processing the additional data from the accelerometers, ViBand may burn twice as much power as a conventional wearable.

Although the wide potential applications of ViBand are quite intriguing, it seems that some security concerns need further exploration. As mentioned in the beginning of this article, a sophisticated algorithm can extract the passwords and PINs from a conventional wearable. Now that ViBand is gathering much more information than conventional devices, the security of the device seems to be even more vulnerable.

In addition, since the device itself is naturally burning more power, it is more likely to employ low-energy communication methods which further raise the security concerns.

The details of this technology are presented at the Association for Computing Machinery’s User Interface Software and Technology (ACM UIST) Symposium in Tokyo.