Why spectral masks?

Written on 3:15 AM by ooe

Crosstalk statistics only give useful characterisation of the frequency domain magnitude of coupling (for two given pairs phase can also be measured repeatably – and some noise cancellation techniques propose doing exactly this. However, it is not correlated with the corresponding measurement on any other pair of pairs). Hence only this is available to the ANFP – which is concerned only with the aspects of injected power after it has experienced a crosstalk coupling. The loss of phase information would mangle any other transform domain into uselessness (including the untransformed time domain).

In principle one could introduce other limits too, such as a limit on total power sent to line. This was not done in the ANFP principally because the worst-case noise environment would not improve significantly with such limits, and supernumerary limits might exclude future uses of a line for no good reason. Such future uses include sharing a line between two systems via splitter filters (e.g. POTS and ADSL on one line).