Independent Component Analysis for Optimal Passive Sonar Signal Detection

The estimation of the Direction Of Arrival (DOA) of signals in a set of sensors has applications in different engineering areas. In submarine acoustics, the DOA estimates for a passive sonar system has the objective to inform the presence of a target (ship) in one determined direction (bearing). In accordance to the bearing's resolution, space interference can occur, which contaminates signals received in adjacent bearings. In this work, the independent component analysis (ICA) is applied for separating the interference sources and obtain optimal target detection in the respective direction. Results are presented for the passive reception of signals from merchant ships using two algorithms, JADE and FASTICA. The good performance achieved for blind separation points out that the proposed method can assist target detection, being used as a preprocessing tool.