TL;DR
This paper introduces a novel bio-inspired active sonar strategy that leverages the natural sound patterns of sperm whales to achieve covert target detection. By using actual sperm whale vocalizations as sonar waveforms and embedding them within natural whale call sequences, the system significantly enhances camouflage, making it difficult for adversaries to distinguish sonar signals from ambient ocean noise.
What Problem Does This Paper Address?
Active sonar systems, while crucial for underwater target detection, inherently risk being detected by adversaries. Traditional methods to improve covertness, such as altering signal parameters (frequency hopping, time hopping) or using low SNR signals, often leave distinct signatures that can be identified. The challenge is to design sonar signals that are both effective for target detection (providing good range and velocity resolution) and highly resistant to interception, effectively blending into the complex underwater acoustic environment. The paper aims to bridge this gap by drawing inspiration from nature's robust solutions.
Methods
The core of the proposed strategy lies in its bio-inspired approach:
- Bio-Inspired Waveform Design: Instead of creating artificial signals, the researchers utilize true sperm whale call pulses as the fundamental sonar waveforms. These natural sounds are chosen for their inherent characteristics, such as excellent range resolution (RR) and large Doppler tolerance (DT), which are crucial for accurate target measurement.
- Range and Velocity Measurement Combination (RVMC): A specific combination of two sperm whale call pulses is designed to measure both the range and velocity of a target. One pulse is used for range estimation, and the pair is used for velocity estimation.
- Camouflage through Embedding: To further enhance covertness, the designed RVMCs are embedded within a true sperm whale call-train. This means the sonar signals are interspersed with natural whale sounds, making the overall transmitted signal appear as a genuine biological vocalization rather than an artificial sonar transmission.
- Signal Preprocessing: Sperm whale call-trains are preprocessed to filter out ocean noise and remove low-energy call pulses, ensuring the quality and clarity of the selected pulses for sonar use. Wavelet-based denoising is employed for noise reduction.
- Correlation-Based Detection: The sonar receiver uses correlation techniques to identify the start and end times of the transmitted signals and their echoes, enabling the calculation of range and velocity.
Datasets
The paper utilizes a recorded sperm whale call-train dataset.
- Sperm Whale Call-Train Data: A 7-minute and 54-second high-quality recording of a sperm whale call-train was used. This data served as the basis for selecting sonar waveforms and for constructing the disguised signal-train. No new benchmark datasets were introduced.
Key Insights and Ideas
- Nature as a Superior Signal Designer: The research highlights that natural biological signals, honed by evolution, possess inherent characteristics (like broad bandwidth and specific temporal structures) that make them excellent candidates for covert sonar waveforms, outperforming traditional man-made signals in camouflage.
- Camouflage by Mimicry, Not Just Low SNR: The strategy focuses on making sonar signals indistinguishable from natural sounds rather than solely relying on low signal-to-noise ratios (SNR), which can limit detection range or efficiency.
- RVMC for Dual Measurement: The clever use of a pair of sperm whale calls (RVMC) allows for simultaneous estimation of both range and velocity, a critical capability for target tracking.
- The Power of Context: Embedding sonar pulses within a natural call-train is a highly effective camouflage technique, as it leverages the natural acoustic environment to mask artificial transmissions.
Key Results
- High Disguised Ability: Spectrogram comparisons and neural network-based classification demonstrated that the constructed sonar signal-train is highly indistinguishable from natural sperm whale call-trains. A neural network classifier achieved 100% recognition probability for classifying the disguised sonar pulses as sperm whale calls (or ocean noise).
- Effective Range and Velocity Estimation: Experimental results showed that the bio-inspired sonar strategy could accurately estimate the range and velocity of an underwater target.
- Performance Comparison: In simulations, the proposed method showed comparable range estimation performance to traditional LFM pulses but significantly superior velocity estimation performance. This is attributed to the sperm whale call pulse's excellent time resolution and Doppler tolerance.
- Trade-off Overcome: The approach aims to overcome the typical trade-off between detection range and covertness, suggesting it can achieve covertness even at high SNRs and potentially improve covertness by reducing SNR for short-range detection.
Limitations and Future Work
- Limited Velocity Resolution for High Speeds: The paper acknowledges that for targets moving at speeds significantly higher than typical AUVs (e.g., > 3.5 m/s), the Doppler tolerance of sperm whale call pulses might limit the accuracy of velocity estimation.
- Energy Consumption: A significant portion of the transmitted signal energy is consumed by the "maskant" (camouflage) call pulses, which could be a practical consideration.
- Future Work: The authors suggest exploring combinations of sperm whale call pulses with human-made manipulations (like adjusting inter-click intervals) to achieve different goals, such as increasing transmission rates, improving localization accuracy, and potentially developing more advanced covert sonar systems. They also aim to address the limitations in measuring high-speed targets.
Summary updated