Nonlinear nonadaptive space-time processing for airborne early warning radar

The paper deals with the cancellation of clutter echoes in modern airborne early warning radar systems equipped with digital beamforming. Performance is compared of the following two-dimensional filters that operate in the angle and Doppler frequency domains: (i) fully adaptive linear filter, (ii) partially adaptive linear filter, (iii) linear filters with displaced phase centre antenna weights and with fixed Chebyshev tapering, and (iv) the recently conceived nonlinear filter which takes the minimum among the outputs of few linear filters that process the same radar signals with properly selected weights. The fully adaptive filter requires the highest computational burden and a large region of homogeneous clutter to estimate the clutter covariance matrix. Filter (ii) alleviates these drawbacks at the expense of a reduced cancellation. Filters (iii) are the simplest but offer limited performance. The nonlinear filter is a compromise solution, since its computational requirement is slightly higher than filter (iii), but much less than (i) and (ii), and its performance is comparable to filters (i) and (ii). Moreover, since it requires a limited form of adaptation, it shows much higher robustness to nonhomogeneous clutter. Further, the filters are compared against the live data recorded by the experimental radar adaptive array flight test equipment of NRL. The set of real data, which relates to the sea-to-ground transition, allows demonstration of the robustness of the new nonlinear detector to nonhomogeneous conditions.