A novel wideband DOA estimator based on Khatri–Rao subspace approach

A novel DOA estimation method for uncorrelated wideband sources named focusing Khatri–Rao subspace method (FKR) is proposed based on coherent signal-subspace method (CSM) and Khatri–Rao (KR) subspace. Compared with the conventional CSM that simply averages the covariance matrices of different frequency bins after focusing, FKR transforms the covariance matrices into a higher dimensional matrix through KR product. This method has three major advantages: (1) it achieves a higher resolution than CSM, (2) the root mean square error of DOA estimation from FKR is smaller than that of CSM when the initial angles are inaccurate and (3) it performs well even when the number of sensors is reduced to about half of the sources. The performance of the FKR method is demonstrated and analyzed through the computer simulations.

► A DOA estimation method is proposed for wideband sources based on KR product. ► The covariance matrices are combined into a higher dimensional matrix. ► This method achieves higher resolution and smaller RMSE. ► This method performs well even if sources are more than sensors.