Target Depth Estimation Using Hull Mounted Active Sonar

High false alarm rates are a problem in anti-submarine warfare in littoral waters using active broadband sonar. Automatic classification algorithms may help combat this problem by filtering out detections due to non-threatening targets. An important feature for classification purposes is knowledge of the target's depth. Using active sonar with vertical beamforming capabilities, the received signal from a target can be used to find an estimate of the target's depth given an initial guess of the target's horizontal distance from the ship, the bottom profile and the sound speed profile. The estimation is done by an optimization algorithm. The algorithm varies relevant parameters and models signals based on these parameters, comparing the modelled signals with the received signal until parameters providing an optimal fit are found. The modelling is based on using a ray tracing procedure to find eigenrays for a candidate target depth, finding vertical arrival angles and arrival times by use of these eigenrays, and synthesizing a signal based on the arrival angles and arrival times. The ray tracing procedure is done numerically using LYBIN, a platform developed by the Norwegian Defence Logistics Organization (NDLO). Three candidate objective functions for comparing recorded signals to modelled signals are presented. The validity of the eigenray finding procedure is confirmed, and results from testing the optimization procedure on synthetic data when applying the different objective functions are presented. The results show that the method produces target depth estimates which are suitable for classification purposes