Multi-component signal denoising using unitary time-frequency transforms

Recently, the analysis of multi-path configurations became a challenging problem for people working in communication or channel characterization fields. In these areas, the received signal is generally expressed as the sum of the transmitted signal and its time-shifted versions. The estimation of the multi-path parameters, typically based on a matched filtering procedure, depends on the noise-level in the received signal. As shown through experimental results, the usual denoising tools are not always well suited for signals arising from multi-path configuration. In this paper, a new denoising method is proposed adapted to the case of underwater multi-path configuration. This method, which preserves the distance between arrivals, leads to a better noise robustness of the matched filtering procedure.