An optimization of the stable tap-adjustment algorithm for the fractionally spaced equalizer

An adaptive algorithm for the long-term stable operation of a fractionally spaced equalizer is presented. First, the authors review the stable tap-adjustment algorithms which add virtual noise to the received signal. Then they derive a condition such that the additional virtual noise does not degrade the residual mean-square error (MSE). Using this condition, they propose an optimum tap-adjustment algorithm which gives the minimum residual MSE under the specified computational complexities. This algorithm has the following special features: the residual MSE is smaller than that of the tap-leakage algorithm if the convergence speeds are the same, and the required computational complexities of the algorithm can be specified arbitrarily.<>