Semi-blind data-selective equalizers for QAM

This paper proposes a new semi-blind data-selective affine-projection algorithm for channel equalization considering that the transmitted symbols belong to an M-QAM constellation. The main novelties of this algorithm are the combination of the affine-projection (AP) and set-membership (SM) concepts with the formulation of an error measure that is related to a square region (in the complex plane) centered at an element of the constellation. Simulations show that the proposed algorithm tracks slow channel variations as fast as a supervised algorithm, requiring less updates and keeping good BER performance.