Eigencanceler: A new space-time interference canceler

The Eigencanceler is proposed as a new alternative technique for adaptive space-time radar beamforming. The Eigencanceler exploits the eigen-structure of the data covariance matrix to optimize the radar operation jointly in the spatial and temporal domains. While the technique is applicable to a variety of problems this paper focuses on adaptivity in the space and doppler domains. It is shown that signals received by the radar are characterized by time bandwidths products much smaller than the dimensionality of the signal space. Consequently, a limited number of eigenvalues contain most of the interference energy caused by clutter and jammers. The corresponding eigenvectors of the space-time covariance matrix contain all the angle of arrival/doppler frequency information required to filter the interferences out. These eigenvectors span the interference subspace, while the remaining eigenvectors span the noise subspace. The Eigencanceler's weight vector is designed orthogonal to the interference subspace, it provides minimum noise enhancement and maintains specified gain at desired directions and doppler frequencies. When compared to conventional space-time processing methods such as the Minimum Variance Beamformer, the Eigencanceler is shown to be superior in depth of cancellation, robustness, speed and computational complexity.