Blind weak signal detection via dictionary learning in time-spreading distortion channels using vector sensors

Area of Science:

• Underwater acoustics
• Signal processing
• Machine learning

Background:

• Underwater environments present challenges for signal detection due to sparse time-spreading distortion (TSD).
• Existing methods struggle with unknown signals and channel characteristics in TSD channels.
• Passive detection is crucial for covert operations and minimizing system complexity.

Purpose of the Study:

• To develop a blind passive signal detection method for underwater TSD channels.
• To estimate and separate unknown signals from unknown channel impulse responses.
• To evaluate the performance of the proposed method against existing techniques.

Main Methods:

• Dictionary learning (DL) algorithm for signal and channel estimation.
• Development of a log-likelihood ratio detector tailored for sparse TSD channels.
• Performance evaluation through simulations and experimental data from vector sensors.

Main Results:

• The proposed DL-based blind passive method successfully estimates and separates signals.
• The log-likelihood ratio detector shows effectiveness in sparse TSD channel conditions.
• Underwater experiments validate the method's superior detection probabilities compared to other approaches.

Conclusions:

• The dictionary learning-based blind passive method is effective for detecting unknown signals in underwater TSD channels.
• This approach offers improved detection performance in challenging acoustic environments.
• The method provides a robust solution for passive signal detection without prior signal knowledge.