Temporal-Spatial Correlation Analysis for Ship-Radiated Noise Based on Random Matrix Theory

Time-domain methods for temporal and spatial correlation estimation often face challenges due to background noise and environmental factors, leading to low time delay resolution. This limitation reduces the accuracy of underwater acoustic source localization and depth estimation. To tackle this issue, this paper introduces a statistical approach to estimate temporal and spatial correlations in ship-radiated noise. Using random matrix theory (RMT), we employ a factor model to determine the number of dynamic factors and the binary correlation structure of residuals from ship-radiated noise. This technique effectively captures the essential characteristics of underwater acoustic signal dynamics. Our proposed RMT-based modeling approach provides a robust analysis of ship-radiated noise which exhibits a wide frequency band and non-Gaussian nature. Results show that this method is highly sensitive and reliable in detecting variations in the temporal and spatial correlation of ship-radiated noise. The proposed method provides a novel insight into improving temporal and spatial correlation estimation and enhancing situation awareness in underwater environments.