Messages From Underwater: Researchers Reveal Two Water-Air Communication Methods

August 08, 2022 by Chantelle Dubois

Back in 2018, MIT broke the water-air wireless communication barrier. Now, a team at KIT has transmitted messages in a novel way from the Titanic.

Air and water are fundamentally different communication mediums, making direct wireless communication between water-based and air-based systems impossible. Underwater sonar or acoustic signals reflect off water's surface, and radio signals quickly fade once they transition from less-dense air to more-dense water. 


Direct water-to-air communication

Direct water-to-air communication has largely been elusive due to the differing properties of water and air as wireless mediums. Image used courtesy of Christine Daniloff/MIT


Typical workarounds involve using an intermediary to translate messages from a sub-surface vehicle to one above the surface (and vice versa), such as buoys. But are there better ways to communicate?


Leveraging Synthetic Video Generation

Professor Alexander Waibel, director of the International Center for Advanced Communication Technologies, recently put his speech translation technology to the test during an OceanGate Inc. expedition to the Titanic memorial site. 

A press release from the Karlsruhe Institute of Technology (KIT) describes the underlying mechanism of the technology, which translates speech recorded underwater to text. The text is transmitted to the surface and reconstructed to the speaker's voice and face in a synthetic video. This method produces the appearance that communication is happening via a video conference. Text requires less bandwidth than video to transmit, making it more suitable for scenarios where communication links are slower and less reliable. 


Video generation produces realistic voice and lip motions during translation

Video generation produces realistic voice and lip motions during translation from one language to another. Image used courtesy of Weibel et al


If KIT's synthetic video experiment was anything like Professor Waibel's past speech translation method (a process called Face-Dubbing++), it's possible that KIT's underwater team translated voice audio to text and sent the text data to an intermediary on the surface. The intermediary would then forward the data the rest of the distance to the above-water team, where it could finally be used to generate a synthetic video of the speaker’s voice and face. 


MIT Breaks the Water-air Barrier

Other researchers are facing the physics of the water-air communication challenge head-on.

Back in 2018, the MIT Media Lab presented Translational Acoustic-RF communication (TARF). TARF is a system with an underwater component that transmits sonar signals, causing vibrations on the water's surface. A highly sensitive TARF receiver can then use radar to detect those signals and decode the information being sent. The system is also intended to work in reverse, providing a line of communication that doesn’t rely on an intermediary.

TARF supports high data rates by using orthogonal frequency-division multiplexing. The radar portion of the transmitter operates in the 30–300 GHz range, which is in line with 5G.


TARF’s water-to-air communication system

TARF’s water-to-air communication system. Screenshot used courtesy of MIT Media Labs


Bits of data can be inferred by the angle of the reflected signal's vibration, such as 100 Hz for 0 and 200 Hz for 1. 

The team behind TARF performed experiments in various water conditions and with natural and unnatural noise generation. Even in these varying conditions, the team could still transmit messages. 

While the surface vibration generated by TARF is very subtle, the frequency of the signals compared to the frequency of other phenomena happening nearby in the water is much larger in magnitude, so it’s possible to filter out. For example, natural waves are approximately 1 Hz. 


Two Novel Methods for Water-air Comms

While water-air wireless communication is still a long way from a scalable solution, these researchers have demonstrated two methods that show promise—whether it's transmitting text data for synthetic video reconstruction or developing highly-sensitive receivers for detecting water surface vibrations.