A hybrid adaptive technique for dynamic channel equalizations

An adaptive gradient technique is applied for the minimization of a hybrid cost function that consists of amplitude-dependent and constellation-dependent terms. A time-varying weighting coefficient is induced in the combined cost function to emphasize/de-emphasize the role of the two terms in determining the gradient during initializations, convergence, and steady state conditions. This coefficient is recursively updated and leads to improved global and local convergence properties. It is shown that the proposed technique is superior to the method in which the cost function involves fixed weighting.