An improved method for blind separation of complex-valued signals via joint diagonalization

Abstract

The problem of blind separation of complexvalued signals via joint diagonalization of a set of nonunitary target matrices is addressed in this paper. An improved blind source separation (BSS) algorithm is developed based on minimization of the Frobenius-norm formulation of the approximate joint diagonalization problem by using a multiplicative update. Such minimization yields a strictly diagonally-dominant updated matrix at each iteration. With relaxing some constraints on the target matrices, the improved BSS algorithm allows for extended applications. The behavior of the improved BSS algorithm is demonstrated by computer simulation results in comparison with some representative BSS algorithms.

Blind Identification for Communication Signals: An Approach Exploiting Sampling Phase Diversity

“…In practice however, an estimate of G can be obtained by application of the well known AIC [60] or MDL [61] detection criteria on the singular values of (25).…”

“…the case of autocorrelation matrix estimates obtained by a finite number of samples), the operation is rather referred to as AJD. The differences between existing AJD algorithms essentially lie in the choice or implementation of an appropriate optimization criterion, the most popular being the minimization of the off-diagonal element amplitudes of the set [22,25]. For convenience, most AJD algorithms assume that the mixing or channel matrix is full-rank and invertible.…”

“…CVFFDIAG [25]. This algorithm represents an adaptation of the popular FFDIAG [29] to complexvalued problems.…”