The DOA of wideband signals estimation by using ESPRIT based on K-means

Abstract

The ESPRIT (Estimation of Signal Parameters via Rotational Invariance Techniques) is proposed for estimating the DOA (Direction-Of-Arrival) of narrowband signal, which uses the rotation invariance of the received data's signal subspace to estimate the DOA. But the ESPRIT algorithm can't be directly used for the wideband signal. The ESPRIT algorithm based on K-means introduces K-means algorithm to the wideband signal processing, which makes the ESPRIT algorithm can be directly used for estimating the wideband signal DOA. This method takes advantage of sub-band decomposition technique in wideband signal, then estimates DOA in each band by using the ESPRIT method, and finally uses the K-means method to get the DOA of broadband signal by clustering the results of each band. The ESPRIT algorithm based on K-means not only has a small measure error and a small amount of computing but also could estimate the DOA without knowing the signal's prior knowledge. Theoretical analysis and simulation show that the procedural validity of the algorithm.