The subarray MVDR beamformer for active littoral sonar systems

In the littoral ocean environment ambient noise makes passive detection difficult, placing increased demands on active sonar. For active sonar systems that use azimuthally omnidirectional transmitters the bottom, surface, shoreline reverberation and clutter can be quite challenging to overcome when using Doppler sensitive waveforms, especially for low velocity targets. Straightforward application of adaptive beamforming to increase resolution for active sonar is typically challenged by snapshot deficiency. This deficiency is caused by the conflicting requirements for high Doppler resolution for target detection and small time windows for building up a full-rank signal covariance estimates. It has been proposed recently to overcome these limitations through the use of reduced beamspace adaptive beamforming. [1] The Subarray MVDR beamformer is a method of adaptive beamforming that reduces the required degrees of freedom by splitting the array into sub arrays which are first conventionally beamformed. The subarray beams are then adaptively combined to form adaptive weights using the subarray centers as the element spacing. This approach can eliminate the snapshot deficiency problem while still outperforming a conventional beamformer in both Doppler and bearing resolution. This method also more robust to errors in the signal model that can lead to signal suppression. The Subarray MVDR beamformer is analyzed in this paper both against both simulated data and against experimental data collected during the GLINT/NGAS11 experiment conducted by the Centre for Maritime Research and Experimentation (CMRE) in 2011. This experiment consisted of towed arrays deployed by AUVs in a multistatic experiment where the detection of returns of Doppler sensitive waveforms from an Echo Repeater towed at low speed was being attempted in the presence heavy shoreline reverberation. Simulation results indicate that the Subarray MVDR beamformer shows promising rejection of interfering signals - hat are not effectively attenuated by conventional beamforming. The application of the Subarray MVDR beamformer to the experimental data then shows that the Doppler spread of the reverberation ridge is significantly reduced, and the bearing resolution significantly improved.