The issue addressed is how closely ideal performance can be approached using a practical erasure declaration mechanism. A description is given of a specific practical erasure declaration mechanism, and the resulting coded system performance is investigated on the digital subscriber loop. Results indicate that, by appropriate choice of system parameters, it is possible to closely approach the ideal performance.>

The authors describe an approach to combatting impulse noise (IN) on digital subscriber loops using short-constraint length convolutional codes and Viterbi decoding operating in hard-decision-and-erasure mode. Attention is restricted to time-compression-multiplexing transmission schemes using bipolar, or alternate-mark inversion (AMI), line coding, and a rather idealized IN model, although the approach and general conclusions are more broadly applicable. They discuss the tradeoffs in choosing the channel encoding rate to yield an optimum compromise between error-control protection and the combined effects of intersymbol interference and crosstalk. Numerical results are demonstrated for selected typical line configurations.>