FRI-Based DOA Estimation of Multi-Beam Data

Direction of arrival (DOA) estimation is a classic issue in source localization. Recently, a robust gridless finite rate of innovation(FRI)-based method has been proposed to handle it, which offers an efficient solution to the DOA applications. However, the classical FRI algorithm processes the measurements collected by each sensor, and subsequently, the storage and computation costs increase with the number of sensors, which does not satisfy the requirements of DOA-related applications. The information of interest, e.g., DOAs, gets encoded in a few multi-beam samples, and the core problem is to decode this information from the finite samples. In this paper, we develop an extended FRI framework to process the multi-beam data from the digital beamforming (DBF) pre-calibration. The lower storage and computation cost of the proposed FRI algorithm is decoupled from the number of sensors. Our algorithm's validity is demonstrated in simulations and real underwater acoustic data from the South China Sea Experiment.