Damped sinusoidal signal reconstruction using higher-order correlations

Abstract

In this paper the reconstruction of deterministic damped sinusoidal signals from a one-dimensional slice of their multiple correlations is analysed. Signal correlations are estimated using a new higher-order correlation estimator, which allows the exponentially damped structure of the signal to be maintained in any horizontal slice of correlations. This characteristic is of utmost importance for the subsequent application of a linear method to estimate the signal parameters and thus reconstruct the signal. Simulations results show that the correlation-based approach gives better reconstructed signals than data-based methods (KT method) when coloured noise contaminates the signal.