What is ISDN?

ISDN is a digital transmission and switching capability of a network. Standards documents assume it is a public network, but switches with the same functions could be installed in private networks. The essence of ISDN is the ability to link your network access line to anyone else’s access line, and to make that connection on demand. "Anyone else" could be: another subscriber to the ISDN, a node provided by the carrier for a service like packet switching, frame relaying, or cell relaying; or a third party that provides some service like voice processing or credit card verification.

A key component of ISDN BRI is the local loop transmission technology, 2B1Q. This stands for "2 Bits per 1 Quaternary," a way of coding two digital bits into each voltage change on the line (baud). The goal achieved with 2B1Q is to eliminate amplifiers and repeaters in the local loop for digital services. With only copper pairs in the outside plant, ISDN service is easier to install and maintain compared to other services (like traditional T-1) that needed loop repeaters. And 2B1Q makes ISDN available to almost all telco customers on the wires now in place.

The result is seen in the low tariffs for BRI access. While most links between access lines are circuit-switched, a secondary ISDN characteristic (used more outside the U.S.) is to switch a user’s packet data on the D channel to some other location on the packet network (which may or may not be connected via ISDN). For packet data service on the D channel, the carrier must have sufficient capacity in the packet handlers that terminate those D channels in central offices. Because that capacity usually isn’t in place, D channel data in the US most often handles only short transactions like credit card verification.

Access Interfaces

Two ISDN services have been widely adopted:
• Basic Rate, which carries up to two "bearer" channels for voice or data and a small signaling data (D) channel (2B+D);
• Primary Rate, based on a T-1 or E-1 interface that carries 23 or 30 voice channels.
• A full DS-0 is devoted to signaling (23B+D or 30B+D).

Unfortunately the two names (basic and primary) have almost the same common meaning. Think of Basic as being closer to a B channel, so it is the smaller (2B+D). Primary is the other one, the larger capacity access Pipe.

Channels

ISDN uses time division multiplexing (TDM) to create multiple channels on each access link between a user’s site and the ISDN switch, usually in a telco central office:

• A "Data Channel" or "D channel" is always present to carry signaling information (call requests, etc.), in the form of packetized messages. The D channel may also carry packets of user data. D channels never carry regular (PCM encoded) voice traffic or circuit-switched user information. A D channel on one interface may be used to control another interface that does not have a D channel.
• "Bearer" channels may be provided to carry user information. Usually there are two bearer channels in a basic rate interface (BRI)/ though one or none is offered by some carriers. ISDN OB+D is being sold for point of sale devices that use the D channel to verify credit cards, etc. There are 23 (U.S.) or 30 (European) B channels in a primary rate interface (PRI).
• Another channel is always needed for control and synchronization of the local loop. This channel occupies 16 kbit/s in a BRI, 8 kbit/s in a T-1 PRI (the framing bits), and a DS-0 in an E-1 (the same DS-0 used to synchronize the standard E-1 service).

This third type of channel is not available to end users directly, but may be affected by user actions. It is often ignored when discussing bandwidth and access line speeds. You too may choose to ignore it with no operational risk at all.

Signaling

To accomplish its circuit switching, ISDN relies on an "out of band" signaling system, based on data messages sent in the Data or D channel part of the access link. These messages are directed by customer premises equipment to the switch, which may take action on parts of a message and forward other parts to other switches in the network. Except for local calls, it takes multiple switches to complete a network connection.

ISDN out-of-band signaling between switches is Signaling System 7 (SS7/ previously known as Common Channel Signaling System, CCSS). It has been used for more than a decade. ISDN switches communicate with each other over SS7/ which is made up of 56 and 64 kbit/s links among packet switches (signal transfer points, STPs) devoted to signaling. Packet handlers in ISDN switches may also direct some user data from the D channel on an access loop to a separate packet data network.

ISDN signaling between the customer and the switch is similar but different: Digital Subscriber Signaling system 1 (DSS1). US phone lines before ISDN used "in-band" or "channel associated" signaling. That is, the signaling is sent over the same channel as the voice. On analog lines, the channel is the wire pair and it carries either dial pulses, for rotary dialing, or "TouchTone" (technically, dual tone multifrequency or DTMF) for "push button" dialing. When a phone goes "off-hook" (to place a call) the switch returns dial tone. Any pulses or DTMF tones detected by the switch during and immediately after dial tone are considered dialing instructions.

Network Interface Demark Points

The CCITT (predecessor to the ITU-TS) created a reference model for ISDN access loops. The model defines certain points between the customers’ equipment and the carrier’s ISDN switch. The same model fits both the basic rate and primary rate interfaces. What happens across each of these Demarcation points is the subject of extensive technical specifications. Even an incomplete library of ISDN-related specifications occupies more than 5 ft of bookshelf. In fact, it is these specifications for functions or the "functional groups" between adjacent demarcation points that define the ISDN.

Functions at the customer premises are assigned to network termination (NT) equipment and/or terminal equipment (TE). The internal structures of the NT and TE are not specified—only the functions they perform and the interfaces to other equipment or the network. Hardware vendors are free to implement NT any way they (or their customers) want. Functional groups may be combined into a single device which might, for example, hide the S/T demark points from the customer.

Isdn Interface Demarcation Points

Each demark point has a specific purpose:
R: Legacy Equipment
The huge amount of pre-ISDN equipment now in operation cannot be changed out immediately. There will be a need to accommodate legacy interfaces for a long time, perhaps indefinitely.

The ISDN model recognizes the need to work with older terminal equipment (TE-2 in ISDN terminology) by offering the R demark. This can be any data or voice interface: RS-232/ analog voice FXS/ V.35/ fax machine, etc. In a sense, these are not part of the ISDN—older CCITT/ ISO/ and ANSI standards cover them.

The R interface is provided by a terminal adapter (TA) that connects on its network side to an S interface of an NT-2. Some TAs may include the NT-2 functions, which allows them to connect to a T interface on an NT-1. Some TAs will include the NT-1 functions as well, presenting a U interface directly to the network.

How TE-2 output is converted to ISDN formats is up to each vendor. But the resulting format sent to the ISDN network is very strictly defined at the other demark points (S/ T/ U)

SIT: Customer Premises

When terminal equipment is called "ISDN," or TE1/ there is usually an S or T interface on it (some have a U interface). The ISDN phone, PBX/ or other customer premises equipment (CPE) may still need an external NT-1 or NT-2 device, or both, between it and the local loop from the carrier.
Outside of the U.S. and Canada, the carrier provides the NT-1, or more. It is the S or T interface that marks the end of the carrier facilities and the start of a customer’s own equipment. It is here that the S/T interface on CPE makes perfect sense—it lets the CPE plug directly into what the carrier provides.
For example, Germany provides an S interface as the standard service. In other countries, it is the T interface that is the demark point. Because of the S interface point, in Europe the Basic Rate Interface (BRI) is known as SO; the Primary Rate Interface (PRI/ 30B+D), as S2.
The ISDN specifications as late as 1988 assumed that all customers would have service provided at the S or T interface, not the U interface. In the U.S., by contrast, the Federal Communications Commission prohibits the local exchange carrier (LEC) by from providing customer premises equipment—not even NT-1—as part of the service. When the customer has to provide NT-1, the interface to the network (demark) becomes the U point.

U: Local Loop

The U interface in the US is the demarcation between the public network and CPE. It is here that the physical layer transmission format is standardized on 2B1Q for the BRI.
2B1Q encoding reduces by half the number of "voltage changes per second" (baud) needed to transmit a given number of bits per second (bit rate). That is, the bit rate is twice the baud rate. Baud rate determines signal attenuation, so lowering the baud rate allows the signal to travel farther on local loop copper wires. The beauty of 2B1Q is that it allows the great majority of existing local loops (up to 18/000 feet long, or 1400 ohm loop resistance) to carry more than twice as much traffic in digital form (BRI) as they carry in analog form. This covers 99% of US local loops.
The PRI in North America is based on the T-1 extended superframe (ESF) format with B8ZS (bipolar with 8-zero substitution). The bit rate and baud rate on the PRI are the same (1.544 M/s); each pulse represents one bit. In this same form at the S/ T/ and U points, two pairs of wire carry 24 channels in the traditional TDM format.

It is ironic that the T-1 signal is more and more often delivered via 2B1Q technology in the form of High-speed Digital Subscriber Loop (HDSL) transceivers, rather than CSUs. As a local loop technology, the U interface with 2B1Q signal format of HDSL offers many advantages for "high speed" data (up to about 750 kbit/s full duplex on a single twisted pair)—even without a real ISDN switched network. HDSL-2 offers the same speeds on a single pair.
Some PTTs use BRI equipment for leased lines of 64 kbit/s. A few US carriers are using the technology to provision higher speed services (T-1 and fractional T-1) without incurring the costs of repeaters in the outside plant. Over short loops, the U interface can carry enough information for compressed video. Some experiments in "dial-a-movie" used ISDN technology.

V: Central Office Connections
There must be some device at the central office to terminate the local loop, generically called the Loop Termination (LT). An LT could be built into the switch. In many digital services (including 56K DDS and T-1) the LT is a specialized device that converts the signal from the loop to a voltage level, pulse shape, and impedance appropriate for distribution within the CO. A T-1 line signal is converted to the DSX-1 format, for example, by an office channel repeater or similar device. In effect, the LT is a CSU.
If the LT for a BRI or PRI is not built into the ISDN switch, then the interface between LT and switch is called the V interface. However, since this point is by definition within a CO and never available to a customer, its detailed specification is outside the scope of this book. Note that this demark point V is not one of the "V series" interfaces defined by CCITT/ITU—those are serial data ports and modems.

Basic Rate Interface (2B+D)
Residential and small business customers find the capacity of the basic rate interface suited to their needs. Two bearer channels provide voice circuits, simultaneous voice and data, or may be combined into a single data channel of 112 or 128 kbit/s. It is common practice to install multiple BRI lines into a location.
Any ISDN service allows for an unrestricted digital connection (any bit pattern, including all Is and all Os) at 64 kbit/s, at least locally. Now that interconnections between carriers are almost all "clear channel," only rarely is a call between networks in the US restricted for various reasons, like a transmission line with a Is density requirement, or the potential for a switch to insert robbed bit signaling into the data stream. These cases limit a user to 56 kbit/s per B channel, rate adapted to the full 64 kbit/s.
In addition to bandwidth for the bearer channels and the D channel, there is also loop overhead for synchronization and testing (maintenance or M channel). Each of the various functions is assigned a channel based on time division multiplexing. Like T-1 multiplexing in the traditional digital hierarchy, the bit flow is organized into frames and superframes. For the BRI/ frame format varies by direction of transmission and demark point.

U Interface of the BRI
In most of the world the ISDN customer never sees the U interface. The carrier provides the NT-1 device so the service is based on the S or T interface.
In the U.S. and Canada, the carrier presents the U interface on an 8-position modular jack on the end of the local loop. Customers furnish an NT-1 that meets the U-interface specifications (also at a modular jack). An Swire (4-pair) cord with two modular plugs connects them. NT-1 may be powered locally on the U connector.
Basic Rate Access (BRA) was designed to operate over existing analog voice-quality local loops. There is supposed to be no need for special pair selection, conditioning (removal of bridged taps), etc., for loops up to about 18/000 ft. long. Loading coils used to shape frequency response on long loops must be removed. The design limit is a signal loss of about 42 dB from the original signal of 13 to 14 dBm.
The ability to send more than two digital channels in place of one analog conversation depends on the way the digital signal is encoded.

‘U’ Interface Modular Connector Pinout
2B1Q Line Coding
The BRI line signal from the network is coded in "2B1Q". Each pair of consecutive bits is coded into one of four (Quaternary) values. "Quat" is shorthand for a voltage level and the two bits it represents. There are four different quats to denote the four possible pairs of bits/ numbered for reference as ±3 and ±1. Their nominal voltages are defined in a 3:1 ratio, specifically at ±2.5 and ±5/6 V at the output of a transmitter.
This format is also called pulse amplitude modulation (PAM) because the size of the pulse conveys as much information as its polarity. The two bits in each quat, from this viewpoint, are the sign (polarity) and magnitude of the transmitted pulse.
An oscilloscope trace won’t show the translation of bits to quats exactly as outlined for two reasons:
The bits (except for synchronization words, see below) are scrambled ("with a 23rd order polynomial," which sounds like a lot) to break up long runs of Is or Os. The receiver deframes, decodes quats to bit pairs, then descrambles to get the original data.
loop current (1 to 20 mA) may flow on the same wire pair, perhaps to power NT-1. Almost always a sealing current is used to prevent corrosion of electrical contacts in the local loop. Pulses are superimposed on the steady current. The receiver looks for the pulses and ignores the loop and sealing currents.
Both the central office and the customer equipment transmit at the same time, full duplex, over a single wire pair. A "hybrid" circuit, similar to the one in an analog telephone, couples both transmitter and receiver to the same wire pair. Loop current passes through this circuit, and may be fed to the rest of the NT-1 for its power.

2B1Q Line Coding
S/T Interface of the BRI
Keep in mind that NT-1 and NT-2 are "functional groups" that need not be separate physical devices. At each interface a functional group will provide bit timing (clocking recovery, based on the signal received from the network); framing (from unique bit patterns); delineation of B and D channels and octet timing (based on framing); D channel access procedures (signaling); and power feeding.

Wiring
Much of the difference between S and T is in the number of devices that may be attached to each.
T is a point-to-point connection, between NT1 and NT-2/ consisting of two balanced ‘interchange circuits’ of a single copper pair each. Polarity of each pair is not significant. NT-2 may be built into CPE and need not be a separate device.
S may be Pt-Pt, but also supports a "passive bus." The physical layer for this bus interface is two twisted pairs of copper, one for transmission in each direction. Because it connects multiple terminal devices in parallel, polarity of each pair must be consistent at each connector. Two additional pairs may provide power and power monitoring .
Polarity of the two other wire pairs must be maintained, in case they are needed to deliver d.c. power. Reverse power polarity indicates the source is the backup or reserve supply and may inhibit some TE functions.
The transmitter of the NT-2 toward the terminals may have several receivers attached. All of the terminals’ transmitters (one per terminal device) connect to the single receiver on the NT-2. There are procedures (see below) at the S point to control access to the B and D channels by multiple terminals.

Multiple TEs at the S Interface ‘S/T’ Interface Modular Connector Pinout
Primary Rate Interface (23B or 30B + D)
Compared to the basic rate interface (BRI)/ the Primary Rate Interface is relatively simple. PRI has the same pulse shape, framing, rate, and other electrical characteristics at the U/ T/ and S reference points. Even the R interface seen by older terminal equipment could be a "plain" T-1 sharing these layer 1 characteristics (for the physical/electrical interface).
A DS-1 Primary Rate Interface (PRI) is divided into TDM channels using standard T-1 frames. The pulse shape is the same as that defined for the T-1 or the DSX-1/ the digital cross-connect found most often in central offices. This shape is essentially a square wave/ nonreturn to zero (NRZ/ or 100% duty cycle) pulse/ with a peak value of 2.4 to 3.45 V at the transmitter. Some 20% overshoot on the leading and trailing edges is tolerated within defined limits.
PRI requires the Extended Superframe (ESF)/ which has an embedded operations channel (EOC) and a CRC-6 for error checking (two more TDM channels) in the framing bits. The older D4 superframe doesn’t have these features. The EOC carries alarm notifications, statistics, and error indications.
On a T-1/ time slot 24 is the D channel, if there is one present on the interface. A signaling messge on the D channel of one T-1 can control a call that passes through up to 19 other T-1 interfaces. That is how an H^ channel of 1.536 Mbit/s is managed.
On an E-1, signaling messages use TS-15 (actually the sixteenth time slot, as they are numbered 0 to 31). This is the same TS occupied by ABCD signaling bits when that form of signaling is used on an E-1. The first time slot (TS-0) carries framing codes and a small amount of other overhead.
Traditional T-1 did not allow more than 15 zeroes in a row. While Is are sent as pulses (of alternating polarity) Os are represented by no pulses sent. The receiver of a string of Os has to count bit time intervals with no clocking information from the sender. This can soon lead to errors. Voice meets the ones density requirement by never generating an all-Os byte; there are no long intervals without a clock reference. Data, however, may contain long strings of Os that are meaningful and must be transmitted faithfully.

T-1/PRI Frame Stucture
You will find different methods to allow unrestricted user data in a time slot.
T-1: the transmitter substitutes a code word for any all-Os octet. Binary 8-zero substitution (B8ZS) changes 00000000 to either 000+-0-+or000-+0+- The polarity for the first pulse in the substituted byte is made the same as the last data pulse, creating the first of two bipolar violations (BPVs). The receiver recognizes these BPVs in the known pattern and restores the 8 zeroes.
E-1: Line coding includes a scrambling step to avoid long strings of zeros.
Serial interface: V.35 and other synchronous interfaces have separate leads for clocking and so can deliver any number of zeroes in a row.
What’s different between the reference points is defined above layer 1/ in the functional groups between those points. At the physical layer, time slot 24 (the last DS-0 channel in the frame) is the same as the other 23, but it is dedicated to the D (data) channel used for signaling. Clock rate for the line signal at S and T must be extracted from the received signal at the U interface. This requirement arises from the time division nature of the transmission—the network switch must operate at the same speed in both directions. And there is no way to adapt slower rate data to fit the B channel except, in some cases, from the R to the S interface.
Unlike the BRI, where NT-1 extracts receive clock and generates transmit clock, "turning the clock around" at a PRI necessarily becomes the responsibility of the NT-2. The terminal equipment, if it has a T-1 interface, is also loop-timed, so it too sends at the same bit rate it receives.
NT-1 on a PRI is transparent to the data, clock, and framing. It acts as a repeater, not a controller, for timing. On equipment that has more than one PRI to the same network, the recommendation is to derive clock from the physical interface that carries the D channel from the CPE to that network.
This improves the chance that signaling will be transmitted successfully. The connector at the demark is an RJ-48/ 8-pole modular jack. The terminal side equipment presents an RJ-48 plug on a cable/ which itself may plug into the terminal. Pinout is given in .
Powering the PRI also differs from the BRI or, more precisely may differ when defined. At this writing, no power is to be applied to the signal leads at the S or T interfaces. The NT-1 at the U interface must not apply power to the loop, but the CO may arrange with the customer to deliver power or loop sealing current.

PRI Reference Points PRI Interface Modular Connector Pinout

<NEXT>