“…A similar behavior is observed in the AXT mitigation methods, in which the rates of some lines achieve greater rates than other [10], [11], [12]. In [10] and [11], the lines are benefited distinctly due to the decision feedback structure, which provides more prediction taps for the last user, thus increasing the AXT mitigation, and, consequently, providing more rates for these lines. Conversely, in [12] the first lines appreciate the higher rates, because in this method the gains of each line are the diagonal elements of the upper-triangular matrix of a QR decomposition [13].…”