Look into the systematic behaviour of insertion loss deviation in data grade channels, its partial measurement and characterisation as a specification requirement

The author states the precise definitions of insertion loss for transmission lines. The importance of the insertion loss deviation (ILD) is elucidated. The author highlights the fact, that ILD has basically two components one which essentially represents an offset of the summed up operational attenuation of the components, and a second component, which is oscillatory in nature over frequency. Both components of the ILD result in signal reflections, arriving overlaid to the signal at the receiver end. Hence, ILD represents a noise, arriving simultaneously to the signal at the receiver (ghosts). There are multiple phase or time delays between the signal and the added reflection components of the ILD. In fact, an added oscillatory component is generated during the travel of the signal in the channel. This component can be therefore considered, for specification purposes only, as being pseudo-stochastic in nature with respect to the globally varying channel configurations. ILD may be, from a noise point of view, comparably restrictive on channel or link performance as NEXT. This is important as this ILD noise as not, to date, been compensated for in the deployed data protocols. Though, it is easily feasible to implement it at the chip level. It may, however, require longer negotiation cycles. Within ISO/IEC prime importance has been so far attached only to the channel performance based upon the attenuation to near-end cross-talk ratio of links and channels. But this is not the only limiting characteristic imposing severe restrictions on the link or channel performance, especially not for the operating frequency ranges considered for class E and F links and channels with inherently occurring impedance mismatches. In fact, the ILD noise is equally limiting, and thus, comparable to the cross-talk noise as an impacting factor for links and channels with respect to their signal-to-noise ratio, and it will have to be specified accordingly. Otherwise the promised extended bandwidth of these links and channels cannot be used by system designers, due to the lack of critical information, and the peak power spectrum will remain limited to lower frequencies. This would be counteractive to the deployment of class E and F channels, and would favour the continued deployment of class D channels.