An alternative algorithm based on subspace rotation invariance techniques for directions-of-arrival estimation using the compressed data pencil

Abstract

In this paper, an alternative algorithm based on subspace rotation invariance techniques is presented for estimating the directions-of-arrival (DOAs) of uncorrelated narrow-band plane waves in noise using the data pencil compressed by the Karhunen–Loève transform (KLT). It is shown that the compressed data pencil preserves the rotation invariance property, and therefore it can be employed by subspace rotation invariance algorithms, e.g., the original ESPRIT, the TLS-ESPRIT, and PRO-ESPRIT, to estimate the DOAs. Due to the reduced-dimension compressed data pencil, the proposed algorithm has a much faster speed in computation than the above ESPRIT algorithms. In addition, computer simulations illustrate that the proposed algorithm exhibits better statistical characteristics than the original ESPRIT and comparable to the PRO-ESPRIT and the TLS-ESPRIT, respectively.