Localization of buried objects in underwater acoustic in presence of phase errors and unknown noise

This paper presents buried objects localization using a towed flexible antenna in presence of unknown noise. We take into account both the reflection and the refraction of wave at water-sediment interface. A new directional vector, which contains the bearing and the range of objects, is used instead of classical plane wave in MUSIC (MUltiple SIgnal Classification). We propose to estimate phase distortions by a robust optimization algorithm which combines DIRECT (DIviding RECTangles) algorithm and spline interpolation. A fourth-order cumulant matrix is proposed to reduce correlated Gaussian noise. A novel iterative denoising algorithm is developed when the noise spectral matrix is one unknown band matrix. The bilinear focusing operator is used to decorrelate the received wideband signals. It is shown via experimental data that this method has a good performance.