Sparse representation approaches to parameter estimation of completely polarized wideband signals

Three sparse representation methods are proposed for estimating the direction-of-arrival (DOA) and polarization parameters of multiple completely polarized wideband signals. The array employed comprises a number of identically oriented vector antennas (VAs). The first two methods are based on spectral segmentation (SSE) and spectral sampling (SSA), both of which require no prior knowledge/estimation of the correlation functions (or power spectral densities) of signals. The third method is based on spectral truncation (STR) which can also be interpreted as the interpolation/reconstruction of signals’ common correlation coefficient. It is computationally simpler than SSE and SSA but requires preliminary estimation of the signals’ common correlation coefficient within a non-zero scalar multiple. Simulation results have validated the efficacy of the three methods.