Transmitting Video Signals Using (Ultra-)Sound Waves

In a previous article, we explained how Red Pitaya boards were integrated in a signal transmission project in Antarctica – a location without any available communication infrastructure – that used HF radio waves to send information. It showed how the FPGAs operated both in the transmitter and receiver part of a transmission relay, generating and receiving the radio signals.

This was certainly an interesting experiment, and a challenging one due to the conditions, but the underlying concept could be called rather conventional. A different project now uses a similar setup, implementing an extremely unusual technique: transmitting video signal via sound… Ultrasound, to be more precise.

In their paper “Ultrasonic Video Transmission Through Solid Metallic Channel”, Xin Huang and Jafar Saniie from the ECASP Research Laboratory at the Illinois Institute of Technology describe how ultrasound waves (or pulses) offer an unconventional method to transmit livestream video over short distances. This technique provides low attenuation over relatively long propagation distances, but also faces signal dispersion and distortion, with inter-symbol interference (ISI) due to the multipath effect. This effect is caused by signal reflection on material boundaries or flaws, leading to degradation of the signal-to-noise ratio (SNR), negatively affecting the communication performance.

When looking at the experimental layout (see Figure 1, below), we find a mirrored configuration between the transmitter and receiver parts. The setup consists of a reconfigurable SDUC (software-defined ultrasonic communication) platform, with a Raspberry Pi for the user interface and Red Pitaya units for video transmission. The versatile STEMlab boards are responsible for video signal transmission, through the analog outputs, as well as the reception, with the second unit’s inputs picking up and sampling the signal. Both main components are battery-powered COTS, with interfacing capacity for a wide range of wireless application environments. The physical parts include an Aluminum Rectangular Bar (ARB), equipped with 2.5-MHz piezoelectric transducers, combined with 60° oblique angle wedges, generating pulses with a length of 1 ms.

Figure 1 Experimental setup

When analyzing the received pulse wave response, we can observe the ultrasonic beam spreading in the wedges and the mode conversion within the bar, leading to a multipath effect (Figure 2a below). Extracting the pulses with significant amplitude and delays leads to the eight multipath wavelets, shown in Figure 2b. The multipath channel model represents a realistic multipath effect, associated with the ARC channel characteristics.

Figure 2 Ultrasonic multipath channel: a) pulse wave response; b) multipath channel model

Contrary to previous research, where reverberation was reduced via time-reversal and pulse-shaping techniques, Huang and Saniie used ODFM for ISI mitigation. This is a bandwidth-efficient communication technique with a high data rate that can carry different orthogonal subcarriers, modulated with a different sign. Its sensitivity to frequency errors in the channel makes it a robust method. The OFDM bit error rate (BER) estimation was conducted, based on the multipath channel model. QAMs with different modulation schemes were used to modulate/demodulate the binary information, with the modulated symbol transmitted with 2K and 8K subcarriers.

Ultrasonic communication at a high bitrate makes real-time video transmission through solid material feasible. However, the limited bandwidth of the ARB channel leads to lower SNR at high bitrates and, consequently, a higher BER of the video transmission. An experiment was performed to evaluate the highest possible transmission rate in an ARB channel, transmitting a 720p video stream through a 25cm aluminum bar (shown in below Figure 3).

A symbol rate of 250k SPS and 64-QAM modulation was implemented to achieve high-quality video with a bitrate of 1 Mbps and a BER of 3.3 x 10^-4. The impacts of six symbol rate settings (20k, 50k, 100k, 250k, 500k, and 1250k symbols per second) were combined for efficient and reliable video transmission, as well as three QAM modulation orders in order to find the optimal bitrate for three different transmission lengths: 25, 40, and 50 cm. The bitrates and modulation orders are shown in the table below:

This experiment is —besides interesting— certainly original, although you may wonder whether this could ever lead to concrete applications. But then again, who would have thought powerline networking could be capable of providing a stable carrier for your router signal at home some years ago? Every time a new application can take advantage of existing installations and infrastructure for an additional purpose, this should be encouraged and welcomed, since it brings along reductions in the consumption of energy and natural resources, taking us one step closer to carbon neutrality, and Red Pitaya won´t say no to giving a helping hand in reaching that goal.