Polarization Difference Smoothing for Direction Finding of Coherent Signals

The subspace-based two-dimensional direction finding with an array of electromagnetic vector sensors for multiple coherent signals amid unknown noise fields is investigated. The noise can be spatially nonuniform and/or correlated. A novel preprocessing method, called polarization difference smoothing (PDS), is proposed. With PDS the unknown noise is removed by using the difference of pairs of electromagnetic component's data correlation matrices, and the coherent signals are decorrelated by summing these difference correlation matrices. Unlike most other existing preprocessing techniques, such as spatial smoothing and forward-backward averaging, PDS processing does not decrease the array aperture and is applicable to arbitrary array geometry. As an example a uniform rectangular array is considered, and a computationally-efficient propagator-based algorithm (PDS-propagator) is derived. Monte Carlo simulations demonstrate that, with an appropriately chosen PDS matrix, the PDS-based eigenstructure algorithms can offer better performance than the polarization smoothing-based (PS) counterparts. Incidently in the presence of external interference noise, the PDS-based algorithms can underperform the PS-based algorithms.