The dimension reduction approach to direction finding of wide-band sources

In this paper, a dimension reduction approach is presented for direction finding of wideband sources. The spatial interpolating function used to design wideband beamformer is applied here to obtain a matrix, which transforms sensor space to reduced dimension beam space (RDBS), where the beam patterns corresponding to all frequencies are similiar. Then, the compound cross spectral density matrix (CSDM) in the RDBS can be got, and the MUSIC algorithm is used to find the directions. This dimension reduction processing simplifies computation and has good performance.