Receiver optimization for dispersive channels employing coded modulation, with application in high rate digital subscriber line transmission

The authors consider decision-feedback sequence estimation of coded modulated signals transmitted over dispersive channels, and examine its application in high-bit-rate (>144-kb/s) transmission over metallic subscriber lines. To mitigate the adverse effect of colored noise, they consider a scheme called noise prediction. Formulas for optimal receiver filters under a minimum mean-squared-error criterion are derived. A key step in the optimization process is the solution of a max-min problem. Because of the complexity in solving this problem, a suboptimal solution is considered which leads to a convenient lower bound on the potential transmission performance. The derived theory is applied to a few example subscriber lines. Numerical results show the potential of substantial signal-to-signal-ratio gain over pure decision-feedback equalization.<<ETX>>