Optimization Algorithm for Delay Estimation Based on Singular Value Decomposition and Improved GCC-PHAT Weighting.

Area of Science:

• Acoustics and Signal Processing
• Array Signal Processing
• Underwater Acoustics

Background:

• Accurate time delay estimation is critical for effective sound source localization.
• Low signal-to-noise ratio (SNR) conditions pose significant challenges for traditional delay estimation methods.
• Existing methods like GCC-PHAT may not provide sufficient accuracy in low SNR environments.

Purpose of the Study:

• To develop an optimized algorithm for improving time delay estimation accuracy under low SNR conditions.
• To enhance the performance of sound source localization by refining time delay estimation.
• To introduce a novel weighting method, GCC-PHAT-ρ, combined with SVD for improved signal processing.

Main Methods:

• Singular Value Decomposition (SVD) for noise reduction and SNR improvement of acoustic signals.
• Application of an improved GCC-PHAT weighting, termed GCC-PHAT-ρ, to cross-correlation functions.
• Generation of the cross-power spectrum and subsequent inverse transformation to obtain the generalized correlation time domain function for delay difference calculation.

Main Results:

• The proposed SVD-GCC-PHAT-ρ algorithm demonstrated a significant improvement in delay estimation accuracy.
• Experimental results showed at least a 37.95% increase in accuracy compared to standard GCC-PHAT and SVD-GCC-PHAT methods.
• The algorithm proved effective in enhancing delay estimation performance in challenging acoustic environments.

Conclusions:

• The SVD-GCC-PHAT-ρ algorithm offers a robust solution for accurate time delay estimation, particularly in low SNR scenarios.
• This optimization significantly benefits sound source localization applications.
• The proposed method represents a substantial advancement in acoustic signal processing techniques.