Superresolution underwater acoustics

Abstract

We explore the application of some recent results in exponential analysis and sparse interpolation to underwater acoustics, in a joint effort from marine engineers and computational mathematicians. The fact that, in practice, the sampling rate used for the recording of an echoed signal is often a multiple of the Nyquist rate, is regarded as a constraint on the cost and performance of echo sounding devices. Here we illustrate how the latter can be overcome, using a new regular sampling scheme, that can even go well below the Nyquist rate. In the numerical examples an average of 20-25 % of the Nyquist rate is amply sufficient to reliably recognize and reconstruct the echo. The sparse sub-Nyquist method under consideration allows to recover from possible aliasing introduced by the subsampling. The technique does not need newly designed hardware, has low computational complexity and uses a small number of samples.