EE Seminar: Single Microphone Ultrasound Source Localization
~~Speaker: Shalom Elkayam
M.Sc. student under the supervision of Prof. Anthony J. Weiss and Dr. Yossi Yovel
Wednesday, June 29th, 2016 at 15:30
Room 011, Kitot Bldg., Faculty of Engineering
Single Microphone Ultrasound Source Localization
Abstract
The Single microphone localization problem is investigated. This work deals with the problem of localization estimation of a sound signal for locating animals, and more specifically localization of bats in nature using only one microphone and several passive reflectors.
Common passive localization systems are equipped with several base stations, intercepting the transmitted signal and estimating the source location. Often, placing several base stations can be expensive and logistically difficult especially for zoological researchers in an outdoor (nature) experiment. Hence, we have developed an affordable; quick assembly system, and developed the algorithms required to solve the problem under the proposed system constraints.
The signal transmitted by the source and the signals reflected by the reflectors (hereinafter dubbed echoes) are not only overlap in both time and frequency domains, but are also not a-priori known to the user (bats transmissions). An estimation problem which we solved using techniques from other fields of science such as computer vision for semi blind estimation of the transmitted signal using observations of superimposed echoes, although they overlap in both time and frequency domains. This novel method, in contrast to existing algorithms, does not require a-priory knowledge of the transmitted signal model, and is not limited to LFM signals.
The proposed system and algorithms were tested by simulations and compared against several lower bounds (CRLB), for which closed-form expressions are derived. Simulation results demonstrate that the proposed algorithms converge to the CRLB. Finally, we test the system and the algorithms in two experiments: one conducted indoors, in an anechoic chamber, while the other was executed outdoors as an actual field test with real bats.
