Adaptive blind separation of convolutive mixtures of independent linear signals

This paper is concerned with the problem of blind separation of independent signals (sources) from their linear convolutive mixtures. The problem consists of recovering the sources up to shaping filters from the observations of MIMO system output. The various signals are assumed to be linear non-Gaussian but not necessarily i.i.d. (independent and identically distributed). Recently an iterative, normalized higher-order cumulant maximization based approach was developed using the fourth-order normalized cumulants of the "beamformed" data. This approach was source-iterative, i.e., the sources were extracted (at each sensor) and cancelled one by one, in the process yielding a decomposition of the given data at each sensor into its independent signal components. In this paper an adaptive implementation of the above approach is developed using a stochastic gradient approach. Some further enhancements including a Wiener filter implementation for signal separation and adaptive filter reinitialization are also provided. Computer simulation examples are presented to illustrate the proposed approach. (C) 1999 Elsevier Science B.V. All rights reserved.