The Application of SliceSparse Representation of Tensor in DirectionofArrival Estimation for Wideband Sources

In this paper, a novel directionofarrival (DOA) method for wideband sources is proposed. First, the received wideband sources are decomposed into several narrowband components by using the narrowband filter. Second, a series of frequency crosscorrelations (FCCS) between adjacent sensors can be calculated by exploiting the decomposed narrowband components. A Toeplitz matrix can be constructed by these FCCS. Perform the same step on all adjacent sensors. A tensor can then be obtained by stacking all constructed Toeplitz matrices. At last, we use the ideal of slicesparse representation (SSR) to sparsely represent this tensor. The corresponding method is also proposed to achieve the sparse reconstitution of this tensor. The DOA estimation of wideband sources is then achieved. In simulation, the proposed method is compared with the conditional methods, and the results show the effectiveness of the proposed method.