An Adaptive All-Pass Filter for Time-Varying Delay Estimation

The focus of this paper is the estimation a delay between two signals. Such problem common in signal processing and particularly challenging when non-stationary nature. Our proposed solution based on an all-pass filter framework comprising elements: time equivalent to filtering can be represented terms ratio finite impulse response (FIR) its reversal. Using these elements, we propose adaptive algorithm with LMS style update that estimates FIR coefficients delay. Specifically, at each step, updates gradient descent then extracts estimate from filter. We validate our synthetic data demonstrating it both accurate capable tracking time-varying delays.