Estimating the Number and Locations of Boundaries in Reverberant Environments with Deep Learning

Abstract

Underwater acoustic environment estimation is a challenging but important task for remote sensing scenarios. Current estimation methods require high signal strength and a solution to the fragile echo labeling problem to be effective. In previous publications, we proposed a general deep learning-based method for two-dimensional environment estimation which outperformed the state-of-the-art, both in simulation and in real-life experimental settings. A limitation of this method was that some prior information had to be provided by the user on the number and locations of the reflective boundaries, and that its neural networks had to be re-trained accordingly for different environments. Utilizing more advanced neural network and time delay estimation techniques, the proposed improved method no longer requires prior knowledge the number of boundaries or their locations, and is able to estimate two-dimensional environments with one or two boundaries. Future work will extend the proposed method to more boundaries and larger-scale environments.

Figures (7)

• Figure 1: General simulation setting: each NLOS arrival yields an ellipse whose common tangents are reflective boundaries.
• Figure 2: A high-SNR COTANS image with two boundaries, with the curves from the respective boundaries properly intersecting at the ground-truth $(\rho,\theta)$ values.
• Figure 3: The BEI corresponding to the COTANS image in Fig. \ref{['fig:two_boundary_cotans']}, with Gaussian pulses superimposed over the ground-truth $(\rho,\theta)$ values.
• Figure 4: A COTANS image with $N=1$, with one set of curves properly intersecting at the ground-truth $(\rho,\theta)$ value, and the other set observed as global errors.
• Figure 5: The BEI of the COTANS image in Fig. \ref{['fig:one_boundary_cotans']}, with a Gaussian pulse superimposed over the ground-truth $(\rho,\theta)$.