"Sound Source Localization with Data and Model Uncertainties Using the EM and Evidential EM Algorithms" (Post-doc 2015 - 2016)
Publications scientifiques au M2P2
Xun Wang, Shahram Khazaie, Dimitri Komatitsch, Pierre Sagaut. Sound-Source Localization in Range-Dependent Shallow-Water Environments Using a Four-Layer Model. IEEE Journal of Oceanic Engineering, Institute of Electrical and Electronics Engineers, 2017, pp.1 - 9. ⟨10.1109/JOE.2017.2775978⟩. ⟨hal-01702364⟩ Plus de détails...
Sound-source localization in shallow water is a difficult task due to the complicated environment, e.g., complex sound-speed profile and irregular water bottom reflections. Full-wave numerical techniques are currently able to accurately simulate the propagation of sound waves in such complex environments. However, the source localization problem, which generally involves a large number of sound propagation calculations, still requires a fast computation of the wave equation, and thus a simplified model is well advised. In this paper, a four-layer model is considered, which is able to approximate a wide range of shallow-water environments, particularly those in summer conditions. More specifically, the medium is assumed to be horizontally stratified and vertically divided into four layers, and the sound speed in each layer is assumed to be constant or varying linearly. Under this assumption, the wave propagation can be rapidly computed via a classical wave number integration method. The main contribution of this paper is to show the suitability of the four-layer model in terms of source localization in a complex (range-dependent) environment. The sound-speed profile is assumed to be vertically irregular and horizontally slowly varying and the bottom is nonflat. In the forward problem, sound propagation in complex underwater environments is simulated via a time-domain full-wave simulation approach called the spectral-element method. The source localization error due to model imprecision is analyzed.
Xun Wang, Shahram Khazaie, Dimitri Komatitsch, Pierre Sagaut. Sound-Source Localization in Range-Dependent Shallow-Water Environments Using a Four-Layer Model. IEEE Journal of Oceanic Engineering, Institute of Electrical and Electronics Engineers, 2017, pp.1 - 9. ⟨10.1109/JOE.2017.2775978⟩. ⟨hal-01702364⟩
Shahram Khazaie, Xun Wang, Pierre Sagaut. Localization of random acoustic sources in an inhomogeneous medium. Journal of Sound and Vibration, Elsevier, 2016, 384, pp.75 - 93. ⟨10.1016/j.jsv.2016.08.004⟩. ⟨hal-01375680⟩ Plus de détails...
In this paper, the localization of a random sound source via different source localization methods is considered, the emphasis being put on the robustness and the accuracy of classical methods in the presence of uncertainties. The sound source position is described by a random variable and the sound propagation medium is assumed to have spatially varying parameters with known values. Two approaches are used for the source identification: time reversal and beamforming. The probability density functions of the random source position are estimated using both methods. The focal spot resolutions of the time reversal estimates are also evaluated. In the numerical simulations, two media with different correlation lengths are investigated to account for two different scattering regimes: one has a correlation length relatively larger than the wavelength and the other has a correlation length comparable to the wavelength. The results show that the required sound propagation time and source estimation robustness highly depend on the ratio between the correlation length and the wavelength. It is observed that source identification methods have different robustness in the presence of uncertainties. Advantages and weaknesses of each method are discussed.
Shahram Khazaie, Xun Wang, Pierre Sagaut. Localization of random acoustic sources in an inhomogeneous medium. Journal of Sound and Vibration, Elsevier, 2016, 384, pp.75 - 93. ⟨10.1016/j.jsv.2016.08.004⟩. ⟨hal-01375680⟩
Stationarity is a key tool in classical time series. In order to analyze the set-valued time series, it must be extended to the set-valued case. In this paper, stationary set-valued time series is defined via DpDp metric of set-valued random variables. Then, estimation methods of expectation and auto-covariance function of stationary set-valued time series are proposed. Unbiasedness and consistency of the expectation estimator and asymptotic unbiasedness of the auto-covariance function estimator are justified. After that, a special case of the set-valued time series, known as interval-valued time series, is considered. Two forecast methods of the stationary interval-valued time series are explicitly presented. Furthermore, the interval-valued time series is contextualized in the Box–Jenkins framework: an interval-valued autoregression model, along with its parameter estimation method, is introduced. Finally, experiments on both simulated and real data are presented to justify the efficiency of the parameters estimation method and the availability of the proposed model.
Xun Wang, Zhongzhan Zhang, Shoumei Li. Set-valued and interval-valued stationary time series. Journal of Multivariate Analysis, Elsevier, 2016, 145, pp.208 - 223. ⟨10.1016/j.jmva.2015.12.010⟩. ⟨hal-01450819⟩