Whistle mimicking underwater acoustic communication with Nyquist constrained autoencoder learned whale waveforms

Area of Science:

• Marine bioacoustics
• Underwater communication systems
• Signal processing

Background:

• Biomimetic underwater acoustic communication offers covert military applications by mimicking marine-mammal whistles.
• Existing whistle-mimicking schemes face a trade-off between concealment and communication reliability.

Purpose of the Study:

• To develop a method that improves both covertness and reliability in underwater acoustic communication by mimicking whale whistles.
• To address the limitations of current modification-type whistle-mimicking techniques.

Main Methods:

• Established a whale-whistle waveform-learning framework using false killer whale whistles.
• Employed an autoencoder to learn Nyquist-constrained symbol waveforms.
• Combined learned waveforms with adaptive symbol durations for energy equalization.

Main Results:

• Achieved the highest Evaluation of Covertness of Sound Mimicking Marine Mammals (ECSM3) score with minimal variation.
• Demonstrated the lowest Bit Error Rate (BER) among same-rate schemes in simulated and real-world underwater conditions.
• Attained a significantly lower BER (0.0245 at 250 bps over 1.5 km) compared to V-DPSK, HODFM, and CV-CFM.

Conclusions:

• Learned whale-like pulses can simultaneously enhance covertness and reliable underwater acoustic communication.
• The proposed framework effectively balances mimicry and communication performance.
• This approach offers a promising solution for covert military operations and other underwater communication needs.