What is DSL?
Types of DSL
Problem With DSL Deployment?
Copper lines have it rough. Twisted pairs get spliced and
respliced. They're subject to a wide variety of vibrating,
radiating, noise-making elements in the environment, not to
mention bad weather if they're aerial, digging cuts and
contamination if they're buried. Loop design parameters such
as working length, cable gauge, and the number and location
of add-on physical components can be expected to affect DSL
performance in a variety of ways. These are some of the most
common factors that disqualify loops for DSL:
Cable length - DSL was designed deliberately to use
low power, to avoid interfering with other transmissions.
But that also means its signals can get lost in the
background noise if they have to travel over too long a
cable. Right now, 12 to 15 thousand feet is the usual
limit, as much as 18 thousand feet under otherwise ideal
Load Coil - is a piece of electrical hardware that
was widely installed in an earlier time to improve voice
transmission by effectively boosting low frequencies,
especially over lines longer than 15 kilofeet. In so
doing, however, it tends to suppress high-frequency
signals, such as DSL. A load coil essentially disqualifies
a loop for DSL service - if you can see that it's there.
Bridged taps - have been included on many loop
lines to increase efficiency in use of the outside plant.
They create multiple signal paths that can pose special
transmission challenges for high-speed transmissions. A
signal can hit the bridged tap, split in half, and go in
both directions as a weaker signal. A signal can also
"reflect" off the end of a bridges tap, creating
electronic echo and inversion effects. It's even possible
for an inverted signal to collide with its own original
such that the two cancel each other at least partially.
Unshielded cable - is typically found only in the
drop from a telephone pole to the customer's premises, but
that's enough to make the line vulnerable to
high-frequency "ingress noise" that can disrupt DSL. A
typical source of ingress noise in originally AM radio.
Binder groups - running on one of many twisted
pairs in one cable, high-frequency transmissions such as
DSL are susceptible to various kinds of "coupling," when
signals jump across a twisted pair or from one pair to
another. ISDN, HDSL, or T1 signals in a binder group are
all capable of coupling into a pair that's carrying DSL
and interfering with the DSL signal.
The presence of any of these factors in a loop's design
is a good clue to disqualification, but loop design does not
guarantee loop performance. Anyway, loop records often don't
match actual loop configurations. Records that start out
accurate can suffer through incomplete updates or simple
data-entry errors. An out-of-date diagram, a typo in the
written records, or an unrecorded backhoe incident of 20
years ago might turn a loop that functions fine for voice
into an DSL booby trap. And even if you can establish the
accuracy of loop records, that accuracy may have a very
short shelf life: It disappears the moment anything in the
record or the loop itself changes. When loop records are
known to be inaccurate or unavailable, loop qualification
may have to rely on heuristic extrapolation, using postal
mileage records to estimate loop length and rules of thumb
to postulate the loop's likely makeup.
The big problem with loop qualification by estimation is
the likelihood of false positives - loops that the records
say should work for DSL, but that in fact will not without
sometimes significant remediation investments. The falseness
of a false-positive qualification is usually not known until
the supposedly loop-qualified customer has been sold the
service and is therefore in a position to be frustrated by
and unsympathetic to the service provider's need to fix
what's wrong. The design-based approaches are expected to
yield false findings (positive or negative, in which you say
no to a loop that actually could support DSL) as often as 15
to 30 percent of the time; the industry's top current DSL
providers have experienced false-finding rates as high as 50
percent. Extrapolating from what it costs to bring up to
part for ISDN, $1500 in 1998, even the lower rate means you
need to allocate another $300 for each new DSL customer you
can sign up based on prequalification.
It's been much smarter, so far, to not even try to sign
up any customer whose loop is the least bit questionable,
even though you've already invested the cost of qualifying
that loop. And if such a customer asks for DSL, it's smarter
to just say "You can't have it" and let the customer's
frustration end there. You're in the very awkward and
unwelcome position of having to try to discourage customers
from buying a service they want and you want to sell them.
You're lopping off chunks of your customer base at both
ends, maybe losing a quarter to half of your potential DSL
market before you've started.
Until now, the only real alternative to this risky
business was to apply the loop-testing technique that many
local exchange carriers use to identify the sources of
imported troubles and analyze other existing conditions in
the loop. But these methods are not intended or very well
suited to predicting future performance of a loop for DSL.
For instance, when measuring loop length, they can't
distinguish a straight loop from one that has bridged taps
©2000 New Edge Network, Inc., All Rights Reserved.