Performance of variable step-size dithered signed-error CMA for blind equalization

Recently a dithered signed-error constant modulus algorithm (DSE-CMA) has been proposed, associated with fractionally spaced equalization, for the purpose of low complexity implementation of constant modulus algorithm (CMA). DSE-CMA has robustness properties closely resembling those of CMA under certain restrictions. As the CMA is slow in achieving its minimum mean squared error, so is the DSE-CMA. In this work, we apply an adaptive step-size instead of a fixed one and then examine the performance of few variable step-size algorithms that result in faster convergence while preserve the low computational complexity and robustness properties of the DSE-CMA algorithm. We also derive the excess mean-squared error in the case of noisy channel to examine the robustness of the algorithms.