A Comparative Study of Multilateration Methods for Single-Source Localization in Distributed Audio

• Main Authors: Sran Kitic, Clement Galutier, Gregory Pallone
• Format: Article
• Language: English
• Published: FRUCT 2020-09-01
• Series: Proceedings of the XXth Conference of Open Innovations Association FRUCT
• Subjects: multilateration; localization; distributed; tdoa; array
• Online Access: https://www.fruct.org/publications/acm27/files/Kit.pdf

In this article we analyze the state-of-the-art in multilateration - the family of localization methods enabled by the range difference observations. These methods are computationally efficient, signal-independent, and flexible with regards to the number of sensing nodes and their spatial arrangement. However, the multilateration problem does not admit a closed-form solution in the general case, and the localization performance is conditioned on the accuracy of range difference estimates. For that reason, we consider a simplified use case where multiple distributed microphones capture the signal coming from a near field sound source, and discuss their robustness to the estimation errors. In addition to surveying the relevant bibliography, we present the results of a small-scale benchmark of few ""mainstream"" multilateration algorithms, based on an in-house Room Impulse Response dataset.