A New Iterative Equalizer based on a Deterministic Annealing Process

We describe a new iterative equalizer able to approach the maximum likelihood sequence estimation (MLSE) performance with manageable complexity. This algorithm takes the general form of a continuation or deterministic annealing process. Starting from an exact formulation of the algorithm, we derive a series of approximations, leading to a final structure which has some similarities with the conventional block-iterative minimum mean square error (MMSE) equalizer