Estimates of maximum throughput for simulated local loops

Given the impulse response of a linear time invariant system, input signals with bounded amplitude are proposed which tend to maximize the (L_∞) distance between any pair of output signals over a given time interval. This signal set is used to compute achievable rates for a random sampling of 100 different impulse responses corresponding to simulated local loops. For amplitude constraints and distance requirements (between outputs) of practical interest, the computed rates are significantly greater than the current ISDN (integrated services digital network) basic access rate. A simple model of far-end crosstalk consisting of two coupled channels is also considered, and a simple signaling scheme is presented. The rate achievable with this scheme is compared with the achievable rate with no cross-coupling. The authors also compare this rate with the rate which would be obtained if the crosstalk signal component is treated as random noise with unknown statistics