Blind equalization of time-varying channels: a deterministic multichannel approach

Blind equalization of rapidly changing multipath channels is important for mobile communications, and is addressed by expanding the time-varying channel over a basis. Time-variation offers diversity and in contrast to time-invariant channels, blind equalization of time-varying channels becomes possible under mild conditions provided that the expansion components can be separated. Multichannel data are needed for the challenging non-separable case where it is shown that unique vector FIR equalizers exist under certain channel co-primeness conditions. Apart from persistence of excitation, no extra restriction is imposed on the input. Order selection methods and blind equalizers are derived directly from the output data. Preliminary simulations are also presented