Dave, Since the 16 ANSI test loops were designed with voiceband impedance in mind, and represent, I believe, what were deemed probably worst case situations at voiceband for various properly designed combinations of end section, bridge tap, and loading. Naturally, only the unloaded examples from the ANSI series of 16 were carried into the xDSL testing. With that genesis, however, and some after-the-fact testing and modeling, I am aware of concerns that the ANSI unloaded test loop designs may not really be the bad actors with a wideband complex impedance, which an xDSL ATU-R would present. A lot of analysis and testing has been pursued in the last couple of years with that specific parameter in mind, examining many more combinations of gauge change and bridge taps at the (varying) impedance of a wideband termination. Some of the newer hypothetical examples (all comply with revised resistance design rules)with multiple bridge taps close to the ATU-R, seem the most challenging, especially with higher (VDSL) bandwidths. Some of the VDSL vendors seem to appreciate this, and have implemented more robust equalization/cancellation for their higher bandwidth products. If I recall correctly, back in 1993, the only serious service application contemplated for ADSL was fixed bit rate video on demand, and its cousins. That should not make a big difference, I agree, but we used really stringent BER criteria for the MPEG video because a block error would freeze the picture or a big tile of the picture if it hit an MPEG 'I' frame. The current internet application is more forgiving, and has been pushed harder than video.
I agree that it is generally accepted that T1 service, either provided by the older AMI repeatered lines, or the newer HDSL modem pairs, is 'unusual' in residential areas. The rub we run into is the city where residential and business use is well mixed -- often multiple dwelling units interspersed with office buildings, but also simple zoning that leaves even single family homes down or across the street from commercial buildings. The telco cable network ends up serving both, obviously, and we see both AMI and HDSL. The actionable effect has been that the AMI-based T1s get identified and rolled to HDSL if they share a cable with residential services, a typical situation except in central business districts. We anticipate no problem with HDSL interference into ADSL at low HDSL density, so it is fortunate that virtually all new T1s not directly on fiber have been provisioned with HDSL for over 8 years. What with the normal churn of T1s, that does not leave a big problem for ADSL make-ready. But it is one of the issues that affects interference management and mitigation in an world of unbundled network elements. The timing issue, with very few ADSL lines actually installed (true!) is to get the interference issues flattened before they become inservice problems requiring urgent mitigation. That means the eventual installation and implementation efforts can focus on forward progress rather than maintenance distractions (that is the motivation, at least.)
I'm glad I was no more definitive, and used the words, "I have not seen studies ..." The Bellcore studies that I remember of the sort you describe were addressing spectral interference in the T1 into xDSL direction. I would not be surprised to learn of complementary testing with the opposite orientation, I just never reviewed any personally. My sense, though, is the spectral interference differences between CAP and DMT were a moot issue; carrier decisions as to which to purchase and install seemed to hinge on other factors, especially product availability, standards compliance, and long-run expectation as to price movement. These business criteria seemed to overshadow the technology religious wars between DMT and CAP, which were still great sport.
Thank you for your generous compliment. |
