SkinSense: Efficient Vibration-based Communications Over Human Body Using Motion Sensors

TL;DR

SkinSense demonstrates that the human hand/skin can serve as a covert, low-bandwidth medium for vibration-based device-to-device communication between a smartphone and a wrist-worn sensor. It employs PWM-driven frequency modulation plus time-based ON/OFF encoding, decoded via spectrogram analysis to reconstruct transmitted bits, and is validated across custom and consumer hardware with 13 participants. The work shows a realistic balance between data rate and reliability, achieving up to $6.6$ bps with BER below $0.1$, while also examining security implications such as acoustic side-channels and potential mitigations. Overall, SkinSense provides a practical, stealthyOut-of-Band channel concept that complements traditional radio links and opens avenues for secure co-location and authentication applications, despite current limitations on consumer hardware and higher BER at elevated bit rates.

Abstract

Recent growth in popularity of mobile and wearable devices has re-ignited the need for reliable and stealthy communication side-channels to enable applications such as secret/PIN sharing, co-location proofs and user authentication. Existing short-range wireless radio technology such as Bluetooth/BLE and NFC, although mature and robust, is prone to eavesdropping, jamming and/or interference, and is not very useful as a covert communication side-channel. This paper designs and implements SkinSense, a vibration-based communication protocol which uses human body/skin as a communication medium to create a low-bandwidth and covert communication channel between user-held mobile and wearable devices. SkinSense employs a novel frequency modulation technique for encoding bits as vibration pulses and a spectrogram-based approach to decode the sensed motion data (corresponding to the encoded vibration pulses) to reconstruct the transmitted bits. SkinSense is comprehensively evaluated for a variety of operational parameters, hardware setups and communication settings by means of data collected from human subject participants. Results from these empirical evaluations demonstrate that SkinSense is able to achieve a stable bandwidth of upto 6.6 bps, with bit error rates below 0.1 in our custom hardware setup, and can be employed as a practical communication side-channel.

Figures (10)

• Figure 1: Architecture of (a) ERM motor and (b) capacitive accelerometer.
• Figure 2: Vibrational energy propagation from a source to a medium.
• Figure 3: SkinSense communication protocol.
• Figure 4: Spectrogram of PWMs ranging from 20 to 100.
• Figure 5: (a) Raw accelerometer signal, spectrogram of the accelerometer signal (b) before filtering, (c) after filtering to remove noise.