Ripple II: Faster Communication through Physical Vibration

Nirupam Roy, Romit Roy ChoudhuryNSDI 2016 · OpenReview

TL;DR

This paper introduces Ripple II, a novel vibratory communication system that achieves significantly higher data rates (30 kbps) than its predecessor by using a microphone as a receiver and employing an OFDM-based physical layer. This advancement opens up new possibilities for secure, RF-free communication modalities in applications like authentication and device-to-device interaction.

What Problem Does This Paper Address?

The paper addresses the challenge of enabling efficient and high-bandwidth data communication using physical vibrations, a modality that offers unique advantages like zero RF radiation and contact-based authentication. While previous work demonstrated the feasibility of vibratory communication, it was limited by low data rates (around 200 bits/s) and the use of accelerometers as receivers, which are sensitive to physical motion but less so to subtle vibrations. The authors aim to overcome these limitations by developing a more robust and faster vibratory communication system.

Methods

Ripple II introduces several key innovations:

  • Microphone as Receiver: Instead of accelerometers, the system now uses microphones to detect vibrations. This significantly increases the potential bandwidth but introduces the challenge of distinguishing physical vibrations from ambient acoustic noise.
  • OFDM Physical Layer: The paper implements an Orthogonal Frequency-Division Multiplexing (OFDM) based physical layer. This is a standard technique in wireless communication but is adapted here for vibratory signals.
  • Symbol Selective Adaptive Noise Filtering (SANF): To tackle the noise issue, SANF is proposed. It leverages two microphones to separate ambient sound from physical vibrations. The technique intelligently identifies and filters out erroneous symbols, improving the signal-to-noise ratio.
  • Proactive Symbol Recovery (PSR) MAC Layer: A novel MAC layer protocol is introduced that uses back-EMF (electromotive force) sensing at the transmitter to detect interference. This allows the transmitter to proactively retransmit symbols rather than waiting for a negative acknowledgment from the receiver, leading to improved efficiency.
  • 2x/4x Interleaving: For signaling rate reduction due to interference, the system uses 2x or 4x interleaving of subcarriers, which implicitly communicates the reduced rate to the receiver.

Datasets

Not discussed in the paper. The authors describe their experimental setup and the types of ambient sounds and vibrations they used for testing, but no formal datasets are mentioned.

Key Insights and Ideas

  • Microphones are surprisingly effective vibratory receivers: Despite being designed for acoustic signals, microphones can effectively capture physical vibrations, offering a much higher bandwidth potential than accelerometers.
  • Ambient noise is the primary challenge: The biggest hurdle in using microphones for vibratory communication is filtering out ambient sounds, which can easily overwhelm the desired vibration signals.
  • Back-EMF sensing enables proactive communication: The ability to sense interference at the transmitter via back-EMF is a game-changer, allowing for proactive retransmissions and more efficient MAC layer protocols.
  • Vibratory communication can be surprisingly robust: Even with significant interference, the proposed SANF and PSR techniques, combined with error correction coding, can achieve respectable data rates.
  • Vibratory communication has niche applications: The paper highlights promising use cases in secure authentication (e.g., finger rings) and device-to-device communication where RF is undesirable or impractical.

Key Results

  • Data Rate: Ripple II achieves a data rate of up to 30 kbps, a significant improvement over the ~200 bits/s of its predecessor.
  • Throughput Gain: The system demonstrates a median throughput gain of around 10% with SANF and 26.6% with the addition of PSR, compared to a basic OFDM receiver.
  • Finger Ring Authentication: Achieved 7.41 Kbps with an 8 KHz bandwidth and QPSK modulation.
  • Tabletop Communication: Demonstrated throughputs around 4 Kbps even at distances of nearly 2 feet.
  • Error Reduction: SANF improves symbol correction efficacy, and PSR's proactive retransmissions further reduce symbol errors, especially under heavy interference.

Limitations and Future Work

  • Interference Sensitivity: While SANF and PSR improve performance, they can falter under specific types of mechanical sound spikes or when interference is very low, making it difficult to detect erroneous symbols.
  • Precision vs. Recall in PSR: The PSR protocol is designed to be conservative, leading to strong precision but weaker recall in detecting all erroneous symbols. This is a trade-off to avoid excessive retransmissions.
  • Tuning Interpolation and Sensing: The authors suggest that there is room for tuning the interpolation scheme and the back-EMF sensing sensitivity in PSR to improve performance.
  • Human Factors: For applications like finger rings, placement and user interaction can influence communication quality.
  • Future Work: The authors mention several areas for future research, including modeling the PHY capacity and MAC layer optimality, exploring deeper signal processing for sound cancellation, investigating combined microphone and accelerometer use, and exploring new use cases and applications for vibratory radios.

Summary updated

BibTeX
@inproceedings{roy2016ripple,
  author = {Nirupam Roy and Romit Roy Choudhury},
  title = {{Ripple II: Faster Communication through Physical Vibration}},
  booktitle = {NSDI},
  year = {2016},
  url = {https://openreview.net/forum?id=27Bn3otip1},
}
Figure