Multi-rate Adaptive Nulling of Moving Interferers
4350 N. Fairfax Drive
Arlington, VA 22203
Abstract A fundamental problem in using adaptive beamforming on large underwater acoustic arrays is that the noise due to multiple moving surface ships is non-stationary. As the number of elements increases, the number of snapshots required for covariance matrix estimation increases. As the array size increases, the travel time across the array increases, thereby increasing the interval between snapshots, and the motion of sources measured in beamwidths increases, thereby decreasing the time available for snapshots. The net result is a severe sample support problem in which significant motion occurs before sufficient snapshots can be obtained for matrix estimation. A new approach to this problem is multirate adaptive beamforming. This approach exploits the fact that the nearby interferers appear to be both the loudest and the most dynamic. The approach involves using a relatively few snapshots to estimate the subspace of the K loudest interferers. These strongest interferers are tracked in this interference subspace. In addition, this information is used to determine a dynamically changing orthogonal space in which the K strongest interferers have been nulled. Slow adaptation is carried out in this orthogonal space looking for weak signals. The theory of this approach is explained, including sample support requirements and effects of motion on eigenvalues. The algorithm is illustrated with simulation. Results from application to sea-test data are presented.
Presentation in PDF format