A simple cumulant based algorithm for blind separation of sources
DOI:
https://doi.org/10.1109/ICATT.1997.1235136Abstract
Blind source separation is an important problem in the sphere of antenna array (AA) processing. A variety of algorithms and criteria has been proposed for the solution of this problem (see e.g. [1-3]). However, even in the case of two-element AA the majority of known algorithms has no guaranteed convergence. In their turn, the algorithms having a guaranteed convergence, as a rule, are complicated in a computation respect.
In this work a new algorithm is suggested, providing a guaranteed solution of two independent sources separation on the base of simple algebraic transforms of the fourth-order cumulants of the signals being measured.
References
Jutten, C.; Heraut, J. Blind separation of source, Part I: An adaptive algorithm based on neuromimetic architecture. Signal Processing, 1991, Vol. 24, No. 1, p. 1-11.
Mansour, A.; Jutten, C. Fourth-order Criteria for Blind Source Separation. IEEE Trans. Signal Processing, 1995, Vol. 43, No. 8, p. 2022-2026.
Yellin, D.; Weinstein, E. Multichannel Signal Separation: Methods and Analysis. IEEE Trans. Signal Processing, 1996, Vol. 44, No. 1, p. 106-118.
