Multiuser processing using blind source separation methods

Multiuser Processing refers to a set of fundamental techniques for the improvement of the performances of the modern wireless communications systems. Most of these techniques have been proposed and studied for specific scenarios, according to the multiple access technology to be employed. The theory and methods of blind source separation (BSS) can provide a more general framework over which multiuser techniques may be investigated and analyzed as a particular case. This is the main idea to be exploited along the present paper. We show how this point of view opens perspectives to handle with important problems of multiuser systems, as signal detection and collision resolution. For this task, an information-theoretic based criterion and algorithm, issued from BSS theory, is proposed together with a new algorithm for estimation of the number of active users in the system. Simulation results are carried out to illustrate the performance of the proposal, compared with other blind and trained solutions.