High-resolution Direction-of-Arrival estimation in SNR and snapshot challenged scenarios using multi-frequency coprime arrays

This paper proposes two high-resolution Direction-of-Arrival (DOA) estimators using coprime sensor arrays (CSA) processing broadband signals. The product processor estimates the broadband spatial power spectral density (PSD) by averaging narrowband spatial PSD estimates. These narrowband PSD estimates are formed by multiplying one CSA subarray scanned response with the complex conjugate of the other. Contrastingly, the min processor estimates the broadband spatial PSD by taking the minimum over all subarray periodograms at all processed frequencies for each bearing. The inverse Fourier transform of the broadband spatial PSD estimates the spatial correlation function, which populates the diagonals of a Toeplitz augmented covariance matrix (ACM). The MUSIC algorithm estimates the source DOAs from this constructed ACM. Combining the CSA narrowband PSD estimates over additional bandwidth reduces the number of snapshots needed to attenuate cross-terms in the spatial PSD estimates, providing processing gains for DOA estimation. The MUSIC pseudo-spectra suggest that the product algorithm performs better in scenarios with more sources than sensors and the min algorithm performs better in scenarios with varying source power levels. Monte Carlo simulations show that the new DOA estimators achieve improved precision over previous broadband CSA DOA estimators in snapshot-challenged scenarios.