Hilbert-Ordering Based Clustering of Complex-Excitations Linear Arrays

The clustering of linear phased arrays (PAs) with complex weights is discussed. By exploiting the locality-preservation property of the Hilbert curve, the dimensionality of the problem at hand has been first reduced, and then, a simple clustering algorithm has been applied to maximize the matching of the radiated pattern with a reference one. Both contiguous and noncontiguous partitions of the Hilbert-ordered list of complex excitations have been evaluated to effectively sample the solution space of clustered solutions. A set of representative results, including reference PAs affording steered pencil and shaped beams, are reported for validation purposes as well as to point out the effectiveness of the proposed approach in comparison with state-of-the-art k-means algorithms as well.