Multi-sensor IF Estimation Based on Time-Frequency and Spatial Filtering

Instantaneous frequency (IF) is an important parameter whose accurate estimation is useful in many applications including source localization, blind source separation and signal detection. We propose a computationally efficient and accurate multi-component IF estimator for multi-sensor recordings. The proposed method iteratively estimates the IF of signal components by first tracking the components based on their angle of arrival information and then removing components through time-frequency (TF) and spatial filtering. Experimental results indicate that the proposed method based on joint TF and spatial filtering leads to accurate IF estimates as compared to a recent method that only performs filtering in the TF domain.