Parametric Models and Algorithms for Direction of Arrival Estimation

Abstract

The ability to selectively focus on desired sounds in noisy environments poses a significant challenge newlinewith broad applications, including smart devices, driver assistance systems, smart homes, video newlineconferencing, drones, and hearing aids. Acoustic source localization involves identifying the position newlineof a sound source amidst various factors like reflections, reverberation, and background noise. While newlineextensively studied, acoustic source localization remains an active area of research due to its diverse newlineapplications. Existing localization algorithms face several challenges that limit their effectiveness and newlinepracticality. These challenges include reliance on narrowband models, computational efficiency, adaptability newlineto non-stationary targets, robustness against noise and reverberations, high-resolution localization, newlineand distinguishing between correlated sources. Overcoming these challenges is crucial for the newlinedevelopment of advanced localization algorithms that enhance accuracy, efficiency, and reliability in newlinepractical scenarios. newlineThis thesis is divided into two main parts. Firstly, a comprehensive performance analysis is conducted newlineto evaluate various localization algorithms using real-world datasets, aiming to gain a deep newlineunderstanding of their capabilities. Secondly, a novel technique called trajectory localization (TL) is newlineproposed, which enables accurate estimation of complex trajectories of multiple moving sources simultaneously, newlineeliminating the need for tracking filters. newlineThe technical contributions of this thesis include experimental validation of existing localization newlinealgorithms and the development of wideband signal models and algorithms on real-world recordings. newlineDeep learning architecture is introduced that incorporates direction of arrival (DOA) derivatives for newlineimproving the temporal continuity of DOA, hence resulting in smoother source trajectories. Next, we newlinedevelop parametric models and algorithms for joint localization and tracking tasks and explore various newlinetrajectory localization algorithms. The effective