Time Delay Estimation for Noise-Like Signals Embedded in Non-Gaussian Noise Using Robust Similarity Measures

In this paper, a task of time delay estimation using two sensors receiving a noise-like wideband signal corrupted by non-Gaussian noise is considered. It is challenging task since an elementary interval of signal/noise mixture observation is limited interval and, due to this, time delay estimates obtained in traditional way of cross-correlation data processing can be abnormal with rather high probability. A novel method of time delay estimation based on robust similarity measure has been proposed recently. Here we further analyze and modify this method showing that robustness of similarity measure has to be adapted to noise properties.