Chapter 17 - Application to telecommunications

Abstract

The increasing development of radio communications has led to the generation of an increasing need for both spectrum monitoring by civilians under state control agencies and radio surveillance by military administrations. The spectrum control problem consists in blindly analyzing all the sources, which are detected in the frequency band of the receiver. For some applications, the number of sources may be high. This is, in particular, the case for HF radio surveillance or VUHF airborne spectrum monitoring over dense urban areas. In such conditions, the spectrum control requires the estimation of many parameters belonging to spectrally overlapping sources. This requires a preprocessing of the data consisting in isolating the different components of the observed signals, i.e. blind source separation (BSS). The two possible BSS approaches are: first approach consists in implementing blind separators of convolutive mixtures of sources, either indirectly, i.e. from the blind identification of all the propagation channels, or directly from the data. The second approach relies on the specificity that most of the propagation channels encountered in practice are temporally specular. A channel is said to be specular in the time domain if it is composed of a (small) finite number of paths with different delays and complex attenuations, i.e. if its impulse response is a finite linear combination of Dirac functions.