Echo Energy Estimation in Active Sonar Using Fast Independent Component Analysis

Abstract

In many underwater applications, it is desirable to separate independent signals according to their sources, allowing targets to be distinguished from self-noise, ambient noise and clutter. The long-term goal of this work is to better detect and model target echo under several location in real-ocean environments, and to develop signal processing techniques for echo energy estimation. This paper addresses echo energy estimation problem of active sonar in a set of sensors. This may be done by measuring a noiseless source signal echoed by a target whose acoustic properties are known. We propose an echo energy estimation method based on the following two stages; One is the blind source separation using an independent component analysis (ICA) to separate the remaining mixture into its independent components. We use the principal component analysis (PCA), as a preprocessor, to increase the input signal-to-noise ratio (SNR) of the succeeding ICA stage and to reduce the sensor dimensionality, and followed by the fast Fourier transform (FFT). As the second, after finding an original target echo signal, the energy estimation solution is newly proposed by considering an inverse procedure of the first stage, where the estimated sonar source is used as input for the pseudo-inverse procedure of the ICA filter combined with PCA. Then, we can estimate noise-free energy information of a target echo, which is compared with conventional beam forming method. The real-ocean recorded data demonstrate the performance of the proposed algorithm.