IDSL is ISDN DSL
, and it's a form of symmetric DSL encapsulated
in the ISDN-like electrical signal format.
(More precisely, the electrical signal format referred to is that of the
American ISDN U interface, ANSI T1.601.
What about IDSL in Europe? We are not sure if there is such a thing, it
may be just an American phenomenon.)
By occupying the ISDN frame bits intended for B1, for B1+B2 or for B1+B2+D,
IDSL runs at 64, 128 or 144 kbps, respectively.
IDSL can be thought of as the lowest end of SDSL.
DSLAM vendors implement and ISPs provide IDSL as a separate service rather than just slow the transceivers on their regular SDSL services down to 144 kbps because encapsulating it in ISDN-like framing and following other aspects of the ISDN U interface standard allows IDSL to go through ISDN U-repeaters and loop carriers — that's a very unusual capability for DSL, most DSL flavors only work on direct copper pairs from the CO to the end user.
In the Open SDSL Connectivity Project we generally do not consider IDSL to
be a flavor of SDSL/2B1Q, instead we consider it a
top level
flavor of symmetric DSL on par with SDSL/2B1Q,
SDSL/CAP and SHDSL.
The reason for this classification is that IDSL is not traditionally served
with the SDSL/2B1Q bitpump with an ISDN
framer built on top of it, instead it is traditionally served with an
ISDN U interface transceiver chip like
MC145572.
The reason why IDSL can be attractive (aside from it being the only DSL flavor available in some unfortunate locations) is because it has apparently escaped the ATM craze. When everyone had jumped on the ATM bandwagon and ATM became the service model in the ADSL standard, the G.shdsl standard and most flavors of SDSL/2B1Q, making life miserable for those of us who prefer the modular approach of separate router and DSU, it appears that most (all?) DSLAM vendors have retained the classic HDLC-framed synchronous serial bit stream model for IDSL.
We don't know if any DSLAM vendor has ever made IDSL/ATM. We hope not, but given the circus/zoo nature of the symmetric DSL world, we can never know for sure. But until and unless we hear about IDSL/ATM, we'll assume that there is no such animal and that all IDSL services are DSU-able.
Since (we assume) all IDSL services are DSU-able and there does not seem
to be a wild proliferation of different flavors as with
SDSL/2B1Q, one would hope that someone would have made
an IDSL DSU or at least a device that can function as one.
Alas, the closest we could find was Adtran ISU 128, an ISDN modem that
can apparently be configured in a leased 128
mode in which the D channel
is ignored and the two B channels are concatenated to form a 128 kbps
leased line. This configuration should be able to handle IDSL-128, but we
couldn't find anything that can handle IDSL-144 in which the D channel bits are
added to the IDSL pipe as well.
In accord with Murphy's Law, IDSL-144 seems to be the only IDSL flavor
obtainable out of Covad.
It thus seems that what we need to build is a versatile IDSL DSU, one that can handle all known flavors of IDSL, both 128 and 144 kbps. The straightforward design would be to take an ISDN U interface transceiver chip and couple it with glue logic that would take B1, B2 and/or D channels per user configuration, concatenate them and present them as a single synchronous serial interface.
Although the functional design of such a gadget wouldn't be terribly complex, especially if the user is willing to accept a gated clock on the sync serial interface rather than a PLLed one, the difficulties are elsewhere:
My favourite ISDN U interface transceiver chip MC145572 appears to have been discontinued, and although I can probably get enough pieces for my own hacking, I'm understandably hesitant to sweat designing a board whose principal component is about to become unobtainium.
Any hardware design is an expense of blood, sweat and tears (I hate the layout step with a passion in particular), and this is especially so with anything involving a transmission line interface (the analog stuff). This stuff is different for every transceiver chip, involves lots of blood, sweat and tears, and is almost certain to run into difficult to obtain components.
An alternative solution I'm considering is if perhaps we can cover IDSL with the same gadget we would like to build for SHDSL based on the M289xx chipset. Although that chipset has its own set of unpleasant issues and is definite overkill for simple IDSL, the idea is to cover as many SDSL flavors as possible with as few different hardware designs as possible.
In the meantime one can play with IDSL using our Hack-o-Rocket platform.