Blind source separation based on combined direction finding and constant modulus properties

Abstract

In the blind separation of instantaneous mixtures of multiple cochannel signals using an antenna array, two types of techniques have been used: based on properties of the array manifold, or based on properties of the signals. A well-studied example of the first type is the ESPRIT algorithm, which exploits the parametric structure of the array manifold. A representative of the second type is ACMA, which gives algebraic expressions for the separation of sources based on their constant modulus property, valid for phase-modulated sources. In this paper, we show that the two properties can be combined into a single algorithm.

Maximum likelihood separation of constant modulus signals

“…We have used , whereas the two other sources were located at 01 ; 1 , and 1 was changed from 4 to 30 at steps of 2 . As was demonstrated in [12], in large separation, the ESPRIT algorithm tends to have better performance than the ACMA in terms of SINR. We show that the MLE outperforms both the ESPRIT and the CM-DOA algorithms over the complete range of separations.…”