Evaluating Acoustic Data Transmission Schemes for Ad-Hoc Communication Between Nearby Smart Devices

TL;DR

The paper presents the first independent, real-device evaluation of acoustic data transmission schemes for nearby smartphone-to-smartphone communication. By systematically sourcing, re-implementing, and evaluating eight schemes across diverse distances, device models, noises, and environments, it reveals substantial reliability gaps at practical throughput levels due to multipath, device heterogeneity, and ambient noise. It also demonstrates the value of open replication artifacts by releasing re-implementations and a large real-recordings dataset. The work argues for rigorous real-world testing, standardized testbeds, and open science practices to bridge the gap between simulations and deployable acoustic communication in IoT and mobile contexts. Overall, the study provides a robust framework and practical guidelines for evaluating and designing more reliable low-bandwidth acoustic links in real-world settings.

Abstract

Acoustic data transmission offers a compelling alternative to Bluetooth and NFC by leveraging the ubiquitous speakers and microphones in smartphones and IoT devices. However, most research in this field relies on simulations or limited on-device testing, which makes the real-world reliability of proposed schemes difficult to assess. We systematically reviewed 31 acoustic communication studies for commodity devices and found that none provided accessible source code. After contacting authors and re-implementing three promising schemes, we assembled a testbed of eight representative acoustic communication systems. Using over 11000 smartphone transmissions in both realistic indoor environments and an anechoic chamber, we provide a systematic and repeatable methodology for evaluating the reliability and generalizability of these schemes under real-world conditions. Our results show that many existing schemes face challenges in practical usage, largely due to severe multipath propagation indoors and varying audio characteristics across device models. To support future research and foster more robust evaluations, we release our re-implementations alongside the first comprehensive dataset of real-world acoustic transmissions. Overall, our findings highlight the importance of rigorous on-device testing and underscore the need for robust design strategies to bridge the gap between simulation results and reliable IoT deployments.

Figures (9)

• Figure 1: Acoustic data transmission model. In our experiments, we play the WAV files from the TX implementation on device $D_{\text{TX}}$ (at volume $V_{\text{TX}}$) and record this on device $D_{\text{RX}}$. The resulting WAV file gets decoded using the RX implementation.
• Figure 2: Responses from authors when we requested implementations for 31 publications on acoustic data transmission. We contacted the first authors and all co-authors. For 13 papers, the provided email addresses were outdated, prompting us to search for current contact details manually. We followed up if there was no reply after three weeks. Of the four implementations received, we evaluated three due to runtime errors in the fourth. The other responding authors reported no longer having access to the implementations, citing reasons such as the lead author departing the university or hardware failures.
• Figure 3: Experimental setup. The left picture shows our office setup, with device $D_\text{TX}$ transmitting to device $D_\text{RX}$ at distance $d$. The middle picture shows our setup in the office hallway for larger distances from 5m to 40m. The right pictures show our setup in the anechoic chamber at two different distances.
• Figure 4: Preliminary testing in a best-case hardware scenario. Shorter bars are better. Schemes are sorted by throughput (see \ref{['tab:schemes']}). The error bars show the standard error. Parameters: $d=50cm$; $D_{\text{TX}}=$ loudspeaker Neumann KH 80 DSP; $D_{\text{RX}}=$ microphone Earthworks M23R; $V_{\text{TX}}=-40dB$ interface gain; quiet office environment; N=240.0 measurements.
• Figure 5: Reliability of acoustic transmissions with varying distances. Smaller TER is better. The anechoic chamber allows for measurements up to a maximum distance of 5m. In the office environment, we were able to measure up to a distance of 40m, but the measurements at 1040m were conducted in the hallway. The distance axis is logarithmic, but we only measured at the indicated discrete distances. The error bars show the standard error. The data points of different schemes are dodged to the side to avoid overlap. Parameters: $D_{\text{TX}}=$ smartphone Pixel 4a; $D_{\text{RX}}=$ smartphone Samsung S20 Ultra; $V_{\text{TX}}=$ volume index 19/25; anechoic chamber (\ref{['fig:distance-anechoic']}) and quiet office environment (\ref{['fig:distance-office']}); N=3740measurements.