Spatial Spectrum Estimation with a Maneuverable Sensor Array in a Dynamic Environment

Abstract

Estimation of a time-varying field is essential for situational awareness in many subject areas. Adaptive processing often assumes both the field is stationary and the array is fixed for multiple observation windows. For passive sonar, highly dynamic scenarios such as high bearing rate sources or underwater maneuvers severely limit the utilization of multiple snapshots. Several models are considered for time-varying fields, and a broadband maximum-likelihood estimator is introduced that is solved with an expectation maximization algorithm using as few as one snapshot. The number of estimated parameters can be reduced for broadband data when information, such as shape, is known about the source temporal spectrum. Cramér-Rao bound analysis is used to understand the effects of temporal spectrum knowledge on broadband processing. An example is given for the flat spectrum case to compare with conventional processing. Another feature of dynamic environments is array motion. Since underwater arrays are often subject to motion, the estimate must consider arbitrary, dynamic array shapes. Platforms such as autonomous underwater vehicles provide mobility but constrain the number of sensors. Exploiting a maneuverable linear array with the new estimate allows for left-right or front-back disambiguation and suppression of spatial grating lobes. Multi-source simulations are used to demonstrate the ability of a short, maneuvering array to reduce array backlobes as well as operate in the spatial grating lobe region.