Acoustic Source Localization and Discrimination in Wireless Sensor Networks


Acoustic source localization is an important problem in many diverse applications such as military surveillance and reconnaissance, underwater acoustics, seismic remote sensing, and environmental and wildlife habitat monitoring. Recently, innovative applications such as smart video-conferencing, audio-video sensor fusion and target tracking have also been proposed to utilize source localization. Traditional acoustic source localization methods were developed for wired sensor networks. In Wireless Sensor Networks (WSNs), collaborative source localization is needed where the objective is to estimate the positions of multiple sources by fusion of observations from multiple sensors. There are two broad classes of methods for collaborative source localization. The first class of approaches, where the estimation is done by fusion of the raw sampled signals, is called signal-based, or signal-level fusion. The second class of approaches, where signal features are extracted from raw data at each sensor and the estimation is done by fusion of the extracted features, is called feature-based, or feature-level fusion. The signal-level fusion methods are not suited for WSNs because they require transmission of the raw signal, which is costly due to limited bandwidth and power. On the other hand, the feature-level fusion methods are appropriate for WSNs due to their lower bandwidth and power requirements.

In this project, we develop a feature-based localization and discrimination approach for multiple harmonic acoustic sources in WSNs. The approach uses acoustic beamform and Power Spectral Density (PSD) data from each sensor as the features for multisensor fusion, localization, and discrimination. We use a graphical model to formulate the problem, and employ maximum likelihood and Bayesian estimation for estimating the position of the sources as well as their fundamental and dominant harmonic frequencies. We present simulation and experimental results for source localization and discrimination, to demonstrate our approach. In our simulations, we also relax the source assumptions, specifically the harmonic and omnidirectional source assumptions, and evaluate the effect on localization accuracy. The experimental results are obtained using motes equipped with microphone arrays and an onboard FPGA for computing the beamform and the PSD.



Manish Kushwaha, Xenofon Koutsoukos, Sandor Szilvasi, "Feature-based Collaborative Localization and Discrimination in Wireless Sensor Networks", ISIF Journal of Advances in Information Fusion. (under review)

Manish Kushwaha, Xenofon Koutsoukos, "A Graphical Model Approach to Source Localization in Wireless Sensor Networks", Technical Report ISIS-09-101, ISIS, Vanderbilt University, 2009. [pdf]

Manish Kushwaha, Xenofon Koutsoukos, Peter Volgyesi and Akos Ledeczi, "Acoustic Source Localization and Discrimination in Urban Environments", In International Conference on Information Fusion, July 2009, pp. 1859--1866. [pdf]



The Monte Carlo estimation code for acoustic source localization and discrimination can be downloaded frome here.



aslt.zip63.54 KB