ABSTRACT
This chapter addresses the problem of discriminating radio sources in the context of cellular mobile wireless digital communications systems. Usually, the sources are discriminated in frequency, time, or code. The spreading codes are chosen to be approximately orthogonal so that the sources present share simultaneously the same frequency bandwidth. The chapter describes several methods that can be applied to solve the blind source separation and ISI cancellation problems that we are considering. The presentation includes, in this order, deterministic or zero-order statistics (ZOS), second-order statistics (SOS), and stochastic maximum likelihood (SML) to infinite-order statistics (IOS) methods. The chapter discusses new solutions to the general source separation problem, based on convex and differential geometry driven optimization techniques. These approaches exploit directly specific geometric features of the data model. The chapter considers the blind channel identification and source separation problems in the context of SDMA wireless communication systems.
