सर्वं ज्ञानप्लवेनैव वृजिनं संतरिष्यसि।
न हि ज्ञानेन सदृशं पवित्रमिह विद्यते॥
(With the boat of knowledge, one can cross over all difficulties. There is nothing as purifying as knowledge.)
Deepika Kumari
New Publication: "Spherical Sector Harmonics Based Acoustic Source Localization Using Circular Array" by Deepika Kumari & Lalan Kumar – Presented at ICASSP 2025 | Published by IEEE
Research
About
Welcome to the Research tab, where I explore the cutting edge of Array Signal Processing. With the increasing power of compact devices, sensor arrays are revolutionizing fields from mobile technology to defense systems. Unlike single-channel methods, arrays capture spatial information, enabling superior interference suppression, tracking, and beamforming capabilities. My work delves into the fundamental aspects of sensor arrays, which are critical for a wide range of applications. In the civilian sphere, this includes acoustic scene analysis, speech separation, source localization, room reverberation measurement, assisted living environments, and noise suppression. For defense, I focus on space-time adaptive processing (STAP) algorithms and the design of space-time beamformers to effectively counter clutter and jamming.
A key highlight of my practical work is the filed patent for a Spherical Sector Microphone Array. My core research focuses on advancing Acoustic Source Localization, Beamforming, Source Separation, Ultrasound Imaging and, Array Development, pushing the boundaries of what's possible in these dynamic domains.
Explore my publications to learn more about my contributions to these vital areas. Key areas of my current research are:
Model functions like pressure/velocity in S2H domain
Plane-wave decomposition of a sound field over a spherical sector microphone array has been explored using a novel S2H basis function defined over the sector. This S2H basis function is the most general basis function that takes the usual SH form when the sector is an entire sphere. To achieve orthonormality of the S2H basis function, the orthogonality of the shifted associated Legendre polynomials and of the scaled exponential function have been established for the same sector. Discontinuity at the boundary has been dealt with by solving the Helmholtz equation over the sector.
However, the output of each vector sensor is a vector consists of the acoustic pressure and acoustic particle velocity information. Hence, my immediate plan is to explore and model the particle velocity over sector.
Source Localization
My research focuses on source localization, a crucial area dedicated to precisely identifying where sounds originate in 3D space. This work has significant practical applications, from enabling self-driving cars to hear approaching emergency vehicles and robots to pinpoint malfunctioning machinery, to enhancing acoustic monitoring for wildlife or security and even improving human-computer interaction with voice assistants. My specific contributions involve developing novel techniques using spherical sector microphone arrays, along with new mathematical framework like Spherical Sector Harmonics (S2H). These innovations allow for more accurate and efficient sound source identification in complex environments. Looking ahead, my research aims to further enhance the robustness of these techniques in noisy and reverberant conditions, integrate them with other sensor technologies, and drive their miniaturization for broader accessibility in diverse applications.
