High-speed Digital Subscriber Loop
The assumption in the standards is that the PRI is provisioned over a traditional T-1 or E-1 transmission system. The T-1 transmitter and receiver in most CPE is designed to the DSX-1 specification. Since DSX was created for use within a central office, the distance limit is about 1250 ft.
T-1 loops normally terminate in a channel service unit (CSU) which acts as a regenerator of the "digital" pulses and performs loopbacks for maintenance. CSUs reliably send and receive data pulses over a distance of about 1 mile (1.6 km) on one twisted pair for each direction (dual simplex transmission). Local loops often are much longer, in which case they need powered repeaters in the outside plant. T-1 is relatively intolerant of wire gauge changes, bridged taps, and sloppy splicing. Pairs must be selected for quality or specially engineered. This proves expensive and slow.
In practice these days, a technology called High-speed Digital Subscriber Loop (HDSL) is increasingly likely to be deployed. Eventually it should replace the traditional T-1 and E-1 lines. HDSL may use the same line coding as the BRI 'U' interface, 2B1Q, but at higher rates. More sensitive receivers (than in T-1 equipment) further improve performance. There are forms of HDSL besides 2B1Q that are available, for example Carrierless Amplitude and Phase modulation (CAP, a modem-like signal). HDSL-2 needs only a single pair.
2B1Q encoding can carry 1.544 Mbit/s for a mile or more on a single pair (full duplex), rather than the 2 pairs for T-1. Reducing the baud rate from 772/000 (1.5 Mbit/s) to 386/000 (half a T-1) more than doubles the maximum loop length. At 260 kbaud (1 /3 a T-1) the distance can double again, reaching more than 5 miles over plain cable, without repeaters. As explained under BRI/ 2B1Q transmissions also tolerate gage changes and bridged taps. Consequently, HDSL is much easier and less expensive to install than T-1.
Currently shipping HDSL equipment may have one to four twisted pairs in the local loop. Each pair carries a 0.5 or 0.75 Mbit/s stream of user data, plus some overhead. At both ends of multiple wire pairs, DS-Os are mapped onto time slots of the PRI U interface, in a T-1 or E-1 frame, after removing the additional overhead from each link. LT and NT-1 see only the DSX-1 framing. The HDSL is transparent.
The same sort of HDSL technology is also sold as a short-haul "T-1 modem" or line driver, over a single pair. This version may be useful at the T and U reference points.

NT and CPE at the PRI
Acronym overload aside it is time to look at the user side of the local loop. The functions between the U interface and the user may be in one device or several (NT-1, NT-2, TA/ terminal). The network sees only what happens at U, so that is where the specifications are concentrated. The devices at the user's site are called, collectively, the CPE (customer premises equipment. ).
The terminal, say an analog phone or non-ISDN digital phone, goes off-hook and dials a number. The CPE interprets the dialed digits and converts them into the correct message format to send to the network over the D channel. Most of the analog to digital conversion will take place in a terminal adapter (TA) function for an analog phone. The remainder of the interworking could be relatively simple, done in a combined NT-1 /NT-2, or perhaps aPBX.

At a PRI/ recall, the S and T formats are the same as the U format and N T-1 is transparent to clocking.

For most users, the important events happen at R/ where existing equipment attaches. The network cares only about U. What happens between R and U is simply 'magic' performed by the CPE vendors.

'Magic' at the PRI

Physical/Electrical Interface

NT-1 must receive bits at U with an accuracy of one error in 10/000/000 bits, or better. The required layer 1 functions for an NT-1 are available in chips from many vendors of merchant semiconductors.
Because of attenuation along the local loop of up to 16.5 dB (and another 1.5 dB in an extension cord) the receiver in the NT-1 must detect pulses that are much smaller than the nominal 3 V sent. To ensure reliability, they are tested to 16.5 dB (with an objective of 18 dB) below 2.25V.
S and T requirements are less stringent. Within a given premises, the attenuation over the cable should be relatively small. For those cases, the DSX-1 specification may apply, with its limitation of about 1250 ft. between devices (half that distance from each device to any cross-connect panel). However, the objective is to reach 3000 ft, which is close to the capability of a CSU. At this writing, Bellcore makes B8ZS only an objective, so S and T could use AMI encoding (without B8ZS).
Standards impose many additional requirements that influence the design of hardware. That is, you must assume that any CPE (NT) certified to meet standards has a certain electrical longitudinal balance, tolerance to jitter and wander, and can handle phase transients. Since the user can have no control over these factors they are not covered here.

The NT-1 and -2 portions of CPE have some housework to do when connected and powered up for the first time:

1. NT-1 synchronizes with the network signal, finds frame alignment, and provides the path for the next step.
   
   
2. CPE issues an initialization request to start up the data link on the D channel. An exchange of messages lets the Stored Program Controlled Switch (SPCS), for example the central office switch, give the Terminal End point Identifiers (TEls) needed by the CPE for addresses; the number of TEls varies but usually is one per B channel.
   
   
3. The SPCS then sends a service profile identification (SPID) to the CPE; again the number may vary from carrier to carrier.
   
   
4. The terminal equipment responds with user and terminal identifications (USID/TID).
   
   
5. The switch confirms with a message telling the terminals they are attached; for a phone, the message says in effect "you are attached and onhook but not in service."
   
   
6. Key sets or ISDN phones with multiple buttons can send a Selected Call Appearance (SCA) to the switch to indicate how the switch should handle incoming calls on each B channel.
   
   
7. With that the terminal is ready, so it sends an IN SERVICE message to the switch, which responds with the same message, and the phone is operational.
   

Note that there may be more than one facility (ISDN PRI line) under the control of one D channel. The second PRI has 1.536 Mbit/s available for an H channel. If there is no redundancy for the D channel, the limit is two PRIs per D channel. With redundancy from a backup D channel on another facility, non-Facility Associated Signaling on one D channel may extend over 478 B channels (20 PRIs).

The more interesting part of the PRI CPE is in the NT-2 ^function group," in the software that handles call control procedures. There are vendors of ISDN software who license the source code to various hardware vendors. Thus the NT products from different manufacturers may be running the same software for signaling and other higher layer functions (management, for example). Common software should improve compatibility.

NT-2 software must distinguish among the various modes/ speeds, and bearer services when they are requested in a call setup message. Voice and audio (modem) services are particularly affected, in different ways.

The public switched telephone network (PSTN) delivers many types of information in the form of call progress tones, recorded announcements, and intercept messages. Even when ISDN signaling has an equivalent D channel message, the user may need to have the audible form. For example, a modem can detect and act on an audible busy signal. Attached to an analog port on the NT/TA/ however, a modem will never see the digital message for a busy signal (and wouldn't know what to make of it anyway).

The NT itself could generate the busy tone for the modem, but the network delivers the information more easily by telling the CPE to cut through the B channel to the modem while the switch sends digitally encoded tone. The CPE easily converts any bit stream to an audible form, without interpretation.
Likewise the ISDN switch in the central office must be configured to anticipate the need to deliver audible information as well as the digital messages. This explains in part the complexity of ordering ISDN lines.

On a pure data line, the ISDN switch might simply send the disconnect message with cause #17/ user busy. But for a voice or audio line it would first send a progress message to tell the NT that "further call progress information may be available inband" or "inband information or appropriate pattern now available," and put the busy tone on the B channel.

PRI Modes and Bearer Services

Ordering ISDN Lines
For a basic phone line (Plain Old Telephone Service or analog POTS) these days you usually don't even get to choose rotary dial or touch tone any more—you get both automatically. ISDN lines are not so simple. There are differences in the proprietary "flavors" of ISDN on the central office switch. Then there are the many options on how the switch provides service features and which features the customer wants (and doesn't want).

Fortunately, most of the difficulty for end users disappeared quietly, invisible compared to the Y2K problem being solved at the same time. Where the worst case used to require the user to pick dozens of parameters to specify ISDN BRI service, the latest ISDN customer premises equipment has a single ISDN order code (IOC) that tells the LEC everthing needed to configure the line.

Even if you don't have the manual (and so don't have the IOC) all you need do is tell the phone company which equipment you have (model number, etc.), and perhaps the application, and they will match it to the specific IOC or profile recommended by the hardware vendor. Your Results May Vary

ISDN access tariffs are all local. ISDN by its nature is local access to the central office: from your serving office your traffic goes on as part of a regular transmission system, almost always over optical fiber in the U.S. What the telephone companies charge for almost identical equipment, features, and functions has an astonishing range.

Like POTS, ISDN may have a business rate and a (usually lower) residential rate. Then again, everyone may be charged the business rate, though a residential tariff is offered in many areas.

What happens in the future can't be known, but as predicted in the first edition of this book the trend is toward lower ISDN pricing relative to analog POTS and especially relative to older digital services. In fact, DDS leased lines and Switched 56 circuit-switched channels are generally going up in price. These last two services provide a subset of what is available from a BRI service, yet involve much more manual effort within the telco to install and maintain. For example, most DDS and Switched 56 provides 56/000 bit/s on a 4-wire local loop. BRI provides more than twice that capacity on a 2-wire loop. With easily deployed capability, ISDN BRI continues to be an important offering for LECs facing competition from independent DSL-based CLECs.

For the sake of their own long-term costs, all telcos will have to migrate customers from leased lines and special services (often provided on custom engineered lines) to a generic digital service, ISDN (or frame relay, or ATM/ or IP, or something newer).

The way to convince users to change services is to show them how to save money for comparable performance or how to increase performance for comparable cost. Thus a BRI, to be attractive, should be less than twice the cost of a POTS line for voice. If so, it would be much cheaper than Switched 56, though this is not always the case as some LECs have charged more to carry an ISDN bit labeled "data" than a bit labeled "voice."

The Federal Communications Commission threw a scare into ISDN fans in early 1995 with a ruling that the subscriber line charge (SLC) be applied to each bearer and data channel, not just the 2-wire or 4-wire "line" that provides the interface to ISDN service. Several RBOCs requested a rescinding of the FCCs order, which was granted, but with reservations that it may be reconsidered in the future.

Originally, the SLC applied to the "line" which, in basic phone service, is a twisted pair of copper wires. That same copper converted to ISDN BRI supports three channels (2B+D). Two pairs are needed for a primary rate interface (PRI/ 23B+D). The SLC/ $3.50 per residential line or $6 per business line and rising, makes a considerable difference in the economics of ISDN service for most users. For T-1 access the difference is $1728 per year.

The SLC was created by the FCC and levied on customers to replace revenue lost by the local exchange carriers when access charges to long distance companies were reduced. The result was a shift of costs away from long distance callers at the expense of those who didn't make many LD calls.

Configuring For Terminal Parameters

With an IOC/ you no longer need plan on serious configuration.
The plug-and-play level of most ISDN products in early 1995 was about the same as add-in cards on IBM PC-AT clones. There were many hardware options to select for the application. Then those same selections had to be reflected in the configuration of the ISDN line from the phone company

Help arrived in the form of parameter sets that can be specified with a single designation—an ISDN Ordering Code, For National ISDN-1 (NI-1) service the many lOCs represent an advance but still reflect the legacy of "you gotta configure it" from earlier ISDN versions. The real advance comes with NI-2: lOCs are defined, for up to two terminals, in the form of a base package (there is only one) plus up to six "feature modules" for functions like forwarding and voicemail. Details follow the discussion of NI-1 lOCs.

But assuming the worst case, and you must face the full configuration task, here are the factors to deal with. They apply on either the BRI or PRI/ though ranges of permitted values could vary; (or example, the number of B channels allowed in a connection.

Number of Channels
Under some tariffs, one version of BRI service is offered for D channel access only: the number of B channels is 0. At the other extreme is a PRI with a D channel that controls a second PRI with 24 B channels. That leads to a maximum of 47 B channels possible.
Default: BRI/ 2B + D PRI, 23B + D

Bearer Services
The ISDN switch is capable of passing a 64 kbit/s bit stream transparently, so why should you have to configure the line for voice or data? Because the "N" in ISDN (the network) may not be transparent.

When a connection request is sent to the network by your terminal equipment, the message specifies what kind of call it is: voice/ data, X.25/ etc.
Knowing a connection carries voice/ the network might apply echo cancellation to the channel or convert the encoding from A-law (North American version) to mu-law (European version) of PCM. Any of these processes would destroy data, so must be avoided on data connections.

A data call that is passed to another carrier, or is directed to a subscriber line that has Switched 56 service rather than ISDN service, must be rate adapted to 56 kbit/s.

When the bearer service is X.25 packet switching, from a Packet Assembler-Disassembler (PAD), the network wants to know in advance what volume of traffic to expect and what specific services the customer wants so the X.25 switch may be prepared in advance.

For calls terminating at your site, the specification of which services are supported allows the switch to reject calls that are incompatible. A request for a packet mode connection makes sense only if the called device can deal with packets. This capability is indicated by configuring the line to support packet mode data when the CPE can receive packets.

Options are: speech, 3.1 kHz audio, circuitmode 64 kbit/s/ circuit-mode 56 kbit/s adapted to 64 kbit/s/ and packet mode.

Directory Number (ISDN Phone Number)
Each ISDN interface must have at least one DN/ but there is no limit on the number that may be assigned. A DM is associated with only one interface. One DN will be the default for the interface.

A basic rate line can have a different phone number for each B channel, or both may be considered a hunt group with a single "directory number." Carriers who anticipate exhausting numbers within an area code will appreciate saving some numbers where they are not needed for additional ISDN B channels.

Your Calling Party Number
Does your equipment have to supply a calling party number with each call request? Default is no, but the line may be configured to require such an information element in the call setup request message. This choice makes sense when many users (many phones) call out on the same interface and you want to identify the individual user (or station) to the network, for billing, determining subscription parameters (like presubscribed IXC), or other purposes.

Screening of the outgoing calling party number, for validity is performed by the network when CPN is sent. If not valid, the network will deliver the number to the called party anyway, but also adds the default DN for your interface. You have a separate choice on whether you want your CPN presented to the called device. Default is yes, to permit ANI to function, but the selection can be changed with an information element in the call request.

Subaddress Information
Each user may have another identification, the subaddress information, in addition to the DN. It may be used at the terminating end of a call to route the connection over a private network or to a specific station. The network wants to know if you plan to send or will accept subaddresses, either yours or the called party's. If you don't want or can't use it/ tell the network not to deliver any.

Calling Number Delivery
Do you want to receive the DN of the calling party? Specify yes or no.
Early Cut Through

Normally the network waits for your CPE to confirm that it is ready to receive information (the called extension is off-hook) before opening the channel from the calling party. But when your CPE offers in-band ringing tone while waiting for an extension to answer, you can configure the line to cut through early. This action allows the caller to hear the call progress information.
There is a separate selection of early cutthrough for (1) ISDN and (2) non-ISDN terminal equipment or private network devices behind the NT.

IXC Presubscription
Each interface can designate an interLATA carrier or InterExchange Carrier (IXC). The network will attempt to route circuit mode calls originating here to this carrier. Packet mode calls are handled differently.

Protocol Compatibility Information
End terminals may have special functions or needs. To negotiate compatibility with the equipment called over the ISDN, a TE may send what is called high level and low level compatibility information in the SETUP message. The network does not interpret this information/ and can't act on it. However the SPCS that first takes the signaling message will verify the size and format of each information element, including the compatibility information.

Number of Terminals
The S interface point will support a passive bus. That is, the terminal-side transmitter on the NT-2 has parallel connections to as many as 8 TE-ls or terminal adapters. The count limit is imposed in part by the signal reduction as each attached TE or TA soaks up some of the transmission power. Too many devices will reduce the signal strength to the point where none of them gets a usable signal.

Under older ISDN practices, carriers would provision up to four SPIDs per BRI/ which are still available under some business line tariffs. National ISDN-1 limited the number of TEs per S interface to two. NI-2 also allows only two terminal configurations on a BRI. For simplicity, and for a lack of "pure" ISDN devices, the number in the US seems unlikely to reach 8, the international standard.

Other local restrictions may apply to "custom" versions of the BRI interface.
Service Profile Identification (SPID)

This is the unique layer-3 identification for each circuit-switched terminal device: phone, fax, computer port, etc. There may be more than one such device associated with an NT-1, NT-2, or B channel. For example, an ISDN TA with two voice ports, a serial data port, and an S/T interface would need unique SPIDs for each port so that someone could dial into them. That is, one SPID/ through user configuration of the TA, would identify a fax port, another SPID the data port, and so on.

The SPID is assigned permanently by the carrier at subscription time, to be unique on a switch. The number may be the 10-digit directory number (DN), if there is only one device or port per B channel, or the DN plus a prefix and/or a suffix. The exact form of the SPID is determined by the carrier and depends on the switch make, the software level (National ISDN stage or earlier software), and local practice.

The National ISDN Council fostered the adoption of a standard SPID format that simplifies assignment. The SPID is the DN (NPA-Nxx-XXXX) plus a four-number suffix to identify a B channel (2 characters) and a specific device on that channel (2 characters). Most equipment then can use a SPID of DN+0101 for most applications.

The SPID traditionally has been programmed manually into the TE and the switch in the CO. The switch uses it when the TE is first installed and initializes the layer-3 link (the TA sends its SPIDs to the switch as a form of identification). Enhancements in NI-2 let an unconfigured TA initialize itself and receive the SPIDs assigned to the line by the LEC—no manual configuration for the end user.

The switch from then on uses the SPID to provide the service (mode) appropriate to that device, even when it shares the DN with another device that needs a different mode. For example/ a phone and a router may share a line and a DN, but have different SPIDs to identify which is calling.

Terminal Endpoint Identifier
Each TE must support a layer-2 terminal end point identifier (TEI) for each logical connection on the interface. TEI is part of the frame address (2nd octet) at the S/T interface. Individual NT-2 devices may select which frames to capture based on a match of a stored TEI with the frame address. The 7-bit field allows 128 values, but 127 is the broadcast address, always active. Of the remainder, 0 to 63 may be allocated manually, for PVC connections to a frame handler (packet switch).
TEI values from 64 to 126 form a pool from which the CPE and network negotiate a value to associate with each new switched connection they set up. There must be at least one directory number (DN) for each TEI.

ISDN Ordering Codes (lOCs) for NI-1
The complexity that up until 1994 was the despair of ISDN customers (as well as carriers and hardware vendors) is almost gone. Telcordia (formerly Bellcore), the National ISDN Council, and the National Institute of Standards and Technology (NIST) run a program with NIUF to predefine configuration sets that specify the values of all the possible parameters to be input into the ISDN switch (the "switch translations"). Once defined, these pre-set configurations may be used by anyone.

Generic configuration sets fit broad categories of equipment in typical applications. Vendors can then build default values into their equipment that will interoperate with an ISDN line configured for the same set. It's close to plugand-play, and really there if the TA has an autoSPID feature.

The standard or "Generic" lOCs are widely implemented, replacing those carriers had made up ad hoc. Bell Atlantic, for example, had lumped all the decisions necessary to support the ProShare video conference system under the label "Intel Blue." Rather, Bell Atlantic had three such sets to match the type of switch (AT&T or Northern Telecom) and whether the ISDN "flavor" is proprietary (AT&T Custom) or National ISDN 1. A European version could be yet another set. Now one of the generic lOCs covers almost any need.

Documentation for terminal equipment should state the code (or codes, depending on application) to use. The installer of the equipment need tell the ISDN carrier simply which IOC is wanted. This should be as brief as a few letters, not the individual values for all of the parameters to be set in the switch.
The idea behind lOCs is to select line configuration sets that fit common applications. A given piece of terminal equipment may require different line capabilities (ordered with different codes) for different situations. A hardware vendor might suggest IOC Capability XX for a general purpose terminal adapter, but Capability YY when the same TA is used exclusively for an intense graphics file transfer or LAN interconnection application.

One reason that ISDN equipment had been relatively expensive was that buyers needed extensive technical support to configure and install it. That effort had to be paid for in the price of the hardware, even when the cost to manufacture was modest. As the need for support droped to the level required for a modem, prices for ISDN CPE likewise dropped. A very functional TA/ router with voice ports is only a few hundred dollars (in 2000).

There are three tracks at for defining configurations in terms of an Ordering Code for NI-1. They are distinguished by whether an IOC is associated with a specific vendor's equipment or is intended for a broad application (generic). If generic, there are lOCs that require strict conformaty (one or two letters) and those that require only compatibility (EZ ISDN and a number/letter).

EZ ISDN allows the TA to ignore some features configured on the line, but the TA must not interfer with operation of the switch. This idea is appealing and deserves to take over entirely the specification of ISDN services and the compatibility of CPE.

Generic Ordering Codes for NI-1
Groups of vendors within the NIUF defined core configurations that they saw as fundamental sets of ISDN line features. Each generic core capability registered was proposed to and considered by multiple organizations. There was consensus that the configurations serve a broad purpose—more than to meet the needs of one vendor's equipment.

However, experience in the field with early lOCs determined that some of the original parameter sets are almost never used. These have been relegated to a status called "archived." This means no new lines will be installed based on them, no new CPE will be certified against them, and nobody will support them.
Each generic "capability" is designated by one or two letters. 'A' through 'Q' were published in the first edition of this book, but more were added. The list may still grow as carriers will support NI-1 for some time.

The Capability represents a fixed configuration for all parameters of line, switch, and service. No variations are supported. In the mid-1990s, for only $100, you could have bought a Bellcore paperback book (SR-3480) that spells out all the arcane commands to program a CO switch for each IOC/ in cookbook fashion, with no background information or even explanations for the acronyms. It takes a book because each brand of switch "translates" these parameters differently. This Bellcore publication also described the process to obtain review, confirmation, and registration of lOCs. Bellcore changed its name to Telcordia when SAIC purchased them from the RBOCs.

Generic NI-1 ISDN Ordering Codes
Proprietary lOCs (NI-1)
Specialized terminal equipment may not fit into any of the generic "Capabilities" defined. New applications may call for unanticipated parameter combinations. Hardware vendors have the option of modifying a core or generic capability by making some (not many) changes. The IOC is then designated by the letter of the underlying generic IOC ('X') plus a number ('n') resulting in the form "Xn."
Changes might be the number of directory numbers (DNs)/ assignment of a feature key to a location on a smart phone, etc. Changes would not be accepted in this format if they affected the basic interface; e.g., the number of B channels.

ISDN Ordering Codes for NI-2
Ordering codes for National ISDN-2 are very different. There is only one basic function package, augmented by six optional feature modules. Each base package applies to one terminal device; there may be one or two basic packages assigned to a BRI line/ which give each terminal access to both B channels. The packages may be different for each terminal.

Unlike NI-1 lOCs with relaxed conformity demands (the EZ-ISDN classes), NI-2 demands full compatibility for certification. That is/ the customer premises equipment certified for NI-2 compatibility makes full use of the features for which it is certified. The only exception is for Calling Number Identification, which need not be supported by data-only terminals. NI-1 lOCs remain valid. Those not archived may be used indefinitely for equipment design and for ordering lines.

Basic NI-2 Package
The minimum set for an NI-2 IOC (called NI2-1) includes one directory number, a terminal equipment identifier (TEI), and one SPID. The terminal may use both B channels if equipped to deal with both at one time. Voice and circuitmode data are included.

With either voice or data incoming calls, the network delivers the calling number and a "redirecting number" (where a call was transfered from) if there is one. This is the equivalent of "caller ID" service on home phones, with added information about a location that forwards or transfers a call using the ISDN network capabilities.

Feature Modules for NI2-1
There are six additional feature sets that can be added individually to either or both basic packages on a BRI line.

• Flexible Calling (FC): 3-way conferencing, etc.


• Call Forwarding (CF): may operate on ring-no answer or when the called SPID is busy.


• Voice Mail (VM): answers calls when terminal equipment doesn't.


• Calling Name (CN): the alphabetic information about the owner of the calling number.


• Additional Call offering (AC): notice that another call is waiting; the caller hears ringing rather than an immediate busy signal.

Packet data (P): in X.25 format.
To name a specific combination of features modules, their initials (in parenthases, above) are added to "N2": thus the full set is called N2FCCFVMCNACP.

Where can this go? Not much further as ISDN is now mature enough to reduce the pressure for more development. One feature glaringly absent is the frame relay bearer service. Frame relay, recall, started as part of ISDN, then became an offering provisioned almost always over dedicated access like a T-1 or 56 kbit/s leased line. The technology exists to allow a terminal device to dial a circuit-switched connection to a frame relay port on a packet switch, as discussed elswwhere in this book. Switched virtual circuits on the FR service would make the switched access more attractive.

<PREV>