ABSTRACT
This chapter presents some central issues concerning the problem of blind source separation (BSS) and extraction (BSE) in the context of deterministic and random sources. First of all, a brief introduction to these problems is provided followed by the classical strategy to solve them using independent component analysis (ICA). In addition to that, a new approach for source separation and extraction based on recurrence quantification analysis is introduced, which is illustrated by a representative set of simulations. The possibility of sending chaotic signals for communication purposes gives rise to some interesting theoretical discussions concerning the employment of classical signal processing techniques in such a context. An intuitive way to introduce the BSS and BSE problems is by means of the cocktail party problem. Consider that in a party hearing several conversations at the same time, and, as a major task, wish to separate or just pay attention at only one speaker.
