Table of Contents
Types of ISDN line
- Outgoing call
- Layer 1 activation
- Service profile id
- Call setup by layer 3
- Incoming call
- Call connection
This tutorial is a response to requests from my engineering colleagues for an introduction to ISDN. They wanted to quickly understand the basics, without having to plough through detailed standards and textbooks. I hope this will help.
It is designed to be read continuously from top to bottom and to be easily printed. Hyperlinks to detailed specifications that can be downloaded for free are provided. Both “European” and “North American” variants of ISDN are described.
The Integrated Services Digital Network (ISDN) is a development of the plain old telephone service (POTS) that enables it to carry data and other traffic as well as voice calls. Instead of using a continuously changing analogue voltage on the line between the network and your house, it uses pulses having one of a few discrete voltage levels to encode a series of digits. This is known as “pulse code modulation” (PCM). It is the original “digital subscriber loop” (DSL) technology.
It is more complicated than the POTS way, but has some big advantages:
- Two simultaneous phone calls can be made (or more on primary rate), using the same pair of wires that your POTS telephone used to connect to. This is achieved by interleaving the data for each call, a technique called “time division multiplexing” (TDM). The phone company doesn’t have to dig up the road to change to ISDN and effectively give you a second line.
- Calls can be connected much more quickly – typically within one second over ISDN, compared with 20 seconds or more over POTS. This is especially important when connecting a home computer to an office network (“wide area networking”) or validating credit card transactions, for example.
- Data can be sent faster (64,000 bits per second in each direction) and more reliably, so data calls can be shorter and therefore cheaper. You don’t need a modem to exchange data between computers, although you will probably need a cheaper “terminal adapter” (TA) or ISDN card instead. There is no modem “training” time to wait for (and perhaps pay for) after the call connects.
- Noise, distortion, echoes and crosstalk become inaudible, because the telephone no longer has to measure an exact analogue value, it only has to decide which of a few discrete voltages is present at any particular instant. In most countries, the “trunk” network between telephone exchanges has already been converted to digital technology, for this reason. ISDN just extends it the “last mile” to your home.
- The digits can represent any data, including faxes, files, web pages, sound, pictures and ordinary voice calls. This is the meaning of “integrated services”.
So why isn’t everyone using ISDN? Some disadvantages are:
- ISDN may not be available in your area, because it is expensive to upgrade a telephone exchange to support it. A “network termination unit” (NTU) may also have to be installed at the user end (not necessary in North America), and any “loading coils” removed from the line. Also, it won’t work if you live more than about 5 km from your local telephone exchange, which affects around 10% of users, depending on location.
- ISDN lines typically cost at least twice as much to install and rent as POTS lines. That’s fair enough if you want to be able to make two simultaneous calls, but makes it expensive if you only need one line.
- In order to make fast data calls, both ends must have digital connections. Most internet service providers (ISPs) already support this. Voice calls can be made from ISDN terminals to ordinary POTS phones without problems.
- Long-distance ISDN data calls may be considerably more expensive than voice calls because they can’t be compressed. ISDN data calls may also be charged by time in places where POTS calls are not normally timed. Voice calls over ISDN typically cost the same as over POTS, however.
- Supplementary services such as “caller display”, “ring back when free” and “charge advice” work differently (usually better) than on POTS lines, but are not always available and may cost extra.
- ISDN terminals often need a local power supply, which can be a problem in emergencies. POTS phones normally take their power from the line.
“Basic rate” ISDN lines connect to the network with a normal twisted pair of copper wires. Data is transmitted simultaneously in both directions (known as “full duplex” operation) using some clever echo-cancelling. The stream of bits is organised into two “bearer” channels (B-channels), each of which can transmit data for a single call at 64,000 bits per second, equivalent to 8,000 bytes per second. This is interleaved with a single 16,000 bit/s “D-channel” on the same pair of wires that carries the signalling information for setting up and clearing down calls, and some extra “overhead” bits to allow synchronisation and monitoring of the line (a total of 192,000 bit/s).
The network end of the line is called the “line termination” (LT) and the user end is called the “network termination” (NT). In Europe, the network termination is usually a box attached to the wall of the customer’s premises and owned by the telephone company that converts the two-wire line from the network (the “U interface”) to four wires (the “S/T interface” or “S-bus”). The S-bus allows up to eight items of “terminal equipment” such as telephones or computers to be connected to one line, called a “point-to-multipoint” configuration. Remember that no more than two calls can be made simultaneously, because there are only two B-channels. The network provides sufficient power (around one watt) to run the NT and a simple ISDN telephone for emergency calls.
In North America, it is common for terminal equipment to connect directly to the network, in a “point-to-point” configuration. This saves the cost of the network termination unit, but restricts the connection to a single physical terminal. American ISDN lines do not normally provide any power.
“Primary rate” ISDN lines are much more expensive, use much higher data speeds and different wiring. European “E1” primary rate lines have 30 B-channels, one D-channel and a synchronisation channel (total 2.048 Mbit/s). American “T1” primary rate lines have 23 B-channels and one D-channel (total 1.544 Mbit/s). Cheaper lines are often available that have some of the B-channels disabled but are otherwise identical.
Primary rate connections are presented to the user on four wires – one pair for each direction. They are usually 120 ohm “balanced” twisted pairs using standard network cable, but older installations may use 75 ohm “unbalanced” coaxial cables. Devices called “baluns” can convert between the two types. The connection from the NTU to the network may use additional pairs or optic fibre, depending on the distance. Primary rate connections are always point-to-point.
Primary rate lines are mainly used to connect to a private branch exchange (PBX) in an office or hotel. The PBX then typically provides many POTS telephone lines or basic rate ISDN lines to the users.
The term “broadband” is a rather loose one, often used to describe any connection faster than primary rate ISDN, or a connection capable of passing video signals. Reasonable quality video can in fact be sent over the public ISDN by grouping together six or more B-channels into “H-channels”, which are sometimes called “broadband” connections. Since these connections are “circuit-switched” like a normal phone call, users have exclusive use of the bandwidth and so the connections are of good quality but expensive.
Higher speed ISDN lines (T2, E2 etc) carrying even more B-channels than primary rate lines are also used within the network and to connect to large companies, and can be considered to be broadband ISDN connections.
In many places, the network is being converted to “asynchronous transfer mode” (ATM). This is essentially a stream of small, fixed-size “cells” that can effectively handle both circuit-switched streaming traffic such as voice and video, and packet-switched traffic such as the internet. This too can be considered to be broadband ISDN.
In the local loop, faster DSL technologies such as ADSL are also capable of transmitting video signals (at least in one direction) and are sometimes called broadband connections. However, they typically connect to packet-switched networks and are not called ISDN.
ISDN is specified by reference to the bottom three layers of the OSI seven layer model, described below. These layers are only conceptual – they are not necessarily physically separate in a particular implementation, but they make the protocol easier to understand. The layers communicate with each other using similarly conceptual messages called “primitives”.
The physical layer electrically connects devices to each other. The higher layers can request a physical connection at any time by sending a PH-ACTIVATION request primitive to layer 1. If not already activated, layer 1 will try to wake up the device at the other end of the line (its “peer”) and synchronise to the stream of bits. If successful, layer 1 will then return a PH-ACTIVATION indication to layer 2. If unsuccessful, a PH-DEACTIVATION indication is returned.
Once activated, a continuous stream of bits to and from the peer will be available on the D-channel and B-channels. There is no error correction at this stage, so line noise will occasionally cause errors. Errors tend to occur in bursts, but an average of one bit error per B-channel per minute would be considered normal. Layer 2 sends data to layer 1 in PH-DATA request primitives and receives data in PH-DATA indication primitives.
The data link layer applies error correction to the stream of D-channel bits, to ensure that the essential “signalling” information used for setting up and clearing down calls is transmitted reliably. It also adds addressing, so that signalling information from several terminals can be sent to and from the network over one shared physical channel. The protocol used is known as the “link access procedure for the D-channel” (LAPD). See Q.921 or ETS 300 125 for more information.
The higher layers can request an error-corrected D-channel connection at any time by sending a DL-ESTABLISH request primitive to layer 2. If successful, a DL-ESTABLISH indication will be returned to layer 3. If unsuccessful, a DL-RELEASE indication is returned. Once a data link has been established, layer 3 can use it to reliably transmit signalling messages to its peer in DL-DATA requests and receive them in DL-DATA indications.
The layer 2 specification only describes the D-channel protocol, because the ISDN network is not usually concerned with the content of the B-channels – that is up to the user. The “V.120” or “LAPB” protocol, which is sometimes used to pass data over a B-channel with error correction and flow control, is very similar to LAPD. Voice calls use “mu-law” encoding on the B-channel in North America and “A-law” encoding everywhere else.
The network layer implements the call setup and cleardown procedures over the D-channel. The terminal tells the network what number to “dial” and what type of connection is required, and the network indicates which B channel to use and when to connect or disconnect from it. See Q.931 or ETS 300 102-1 for more information.
The interface from layer 3 to higher layers and the user depends on the application. A standard known as CAPI (common ISDN application programming interface) is often used.
A walkthrough of a typical call setup and cleardown may be the quickest way to understand how the process works. Here is an example of a call from a simple ISDN telephone, starting right from the beginning.
The first user (the “A-party”) picks up the handset of the telephone (the “terminal”). The software inside the telephone signals this event to its layer 3 entity. The layer 3 entity constructs a SETUP message containing a “bearer capability” information element indicating call type “speech”. It needs to send this message through the D-channel to the layer 3 entity in the network (its “peer”), but first layer 2 must be established.
Layer 3 knows that layer 2 is not already established, because it has never received a DL-ESTABLISH indication message from layer 2, in this example. Therefore, it sends a DL-ESTABLISH request message to layer 2 to request establishment of the data link.
We assume layer 2 is in the TEI UNASSIGNED state when it first receives this message. This means it needs to get a terminal endpoint identifier (TEI) from the network before it can establish the data link. The TEI is an address byte which is used in all messages between this particular terminal and the network.
However, layer 2 can do nothing until layer 1 is activated, so it first sends a PH-ACTIVATION request to layer 1 to request activation of the physical link.
Layer 1 is initially in the DEACTIVATED state, which means that there is no signal on the line, a state known as an “INFO 0” signal. The terminal starts transmitting an INFO 1 wake-up signal (any non-zero signal) to begin the activation sequence. The network detects this, and responds with a signal known as an INFO 2, similar to normal communication but with the “activation bit” not set. The terminal synchronises to this signal, and responds with an INFO 3, the normal communication signal from the terminal side. The network detects this, and responds with an INFO 4, the normal signal with activation bit set.
This situation, with the terminal sending INFO 3 and the network sending INFO 4, is the “ACTIVATED” state. The layer 1 entities at each end of the line send PH-ACTIVATION indications to their respective layer 2 entities when this state is achieved.
A full-duplex communication path now exists between the first terminal and the network, consisting of two B-channels and a D-channel (possibly shared with other terminals on the same line).
After receiving the PH-ACTIVATION indication primitive, layer 2 in the terminal transmits a message called an IDENTITY REQUEST to the network over the activated D-channel. The network should respond with an IDENTITY ASSIGN message containing a TEI that will be used in all future messages between this particular terminal and the network. This automatic TEI assignment procedure is only necessary on basic rate point-to-multipoint connections, because on a point-to-point connection there is only one terminal and so the addressing is unambiguous, and a fixed TEI (zero) is used.
To complete the data link establishment process, layer 2 sends a “set asynchronous balanced mode extended” (SABME) command to the network, using the TEI it has just been assigned. The network responds with an “unnumbered acknowledgement” (UA) message to the same TEI. Both layer 2 entities are now established and send DL-ESTABLISH indication primitives to their respective layer 3 entities.
It is quite common for layer 2 to remain established even though no calls are in progress. However, either the network or the terminal may release layer 2 by sending a layer 2 DISC message (acknowledged with a UA), provided no layer 3 calls are in progress. If no layer 2 entities are established, the network may then deactivate layer 1, which saves power. Some countries charge a small fee whenever layer 2 is established, for this reason.
An error-free connection now exists through the D-channel between the network and this particular terminal.
In North America, basic rate terminals are usually required to send a service profile identification (SPID) to the network at this stage, before any incoming or outgoing calls can be made. The SPID is specific to the line and to the type of terminal, and must be obtained from the service provider and configured in the terminal by the user. It typically consists of the phone number of the line with some extra digits such as “0101” added. A different SPID (and TEI) is needed for each simultaneous call, so there are usually two SPIDs per line.
The exact message used to send a SPID on the D-channel varies with the manufacturer of the switch, so information about the switch type must also be obtained from the service provider and configured in the user’s terminal. Typical choices are “Nortel DMS100”, “AT&T 5ESS”, “NI-1” or “NI-2”.
An example of the signalling procedure is described in Q.932 Annex A. The terminal sends an INFORMATION message containing a service profile identification information element to the network using the “dummy” call reference. The network responds with an INFORMATION message containing an “endpoint identifier” information element with a “user service identifier” and “terminal identifier”.
Outside North America none of this service profile identification is necessary, so you can typically connect a terminal to an ISDN line and make calls without further configuration.
Layer 3 in the terminal is at last able to send its SETUP message. It puts the message in a DL-DATA request to layer 2, which wraps the message in a layer 2 information frame (I-frame) and then forwards it to layer 1 in a PH-DATA request. Layer 1 then transmits the frame to the network via the D-channel.
The layer 1 entity in the network receives the frame, removes the “zero bit stuffing” and “flags”, checks the frame length and “frame check sequence”, and passes the frame to layer 2 in a PH-DATA indication. The layer 2 entity checks the address, frame type and sequence numbers, extracts the information field and passes it to layer 3 in a DL-DATA indication – unless the frame arrives corrupted, in which case layer 2 retransmits it automatically until it is received intact.
The network will decode the SETUP message and usually return a SETUP ACKNOWLEDGE message, containing a “channel identification” information element indicating “B-channel 1” or “B-channel 2”. The network will usually also transmit dialtone on the indicated B-channel. The terminal receives the message and connects a “CODEC” to the indicated B-channel to decode the digital information into analogue form so that the user can hear the dialtone.
Everything described above typically occurs within a few tenths of a second, so dialtone is probably already present by the time the handset reaches the user’s ear.
The user then dials the digits of the number to call. As each digit is dialled, the telephone sends it to the network in the “called party number” information element (Europe) or “keypad facility” information element (North America) in an INFORMATION message. When the network has received enough digits to route the call, it typically returns a CALL PROCEEDING message to the terminal. The network then passes the SETUP message to the remote user.
An alternative procedure to this so-called “overlap sending” procedure can be used if all the digits to be dialled are known before the start of the call. In that case, all the digits can be included “en-bloc” in the first SETUP message and the network does not need to return dialtone or a SETUP ACKNOWLEDGE message.
The call is now proceeding through the network.
At the remote end (the “B-party”), the call arrives as an incoming call. The network allocates a free B-channel, activates layer 1 and sends the SETUP to the user’s terminal. If the line is configured at the network as “point-to-multipoint” (basic rate only), the SETUP message is “broadcast” to all terminals on that line using TEI 127. On a point-to-point line, the message is sent using fixed TEI 0.
The ISDN terminal(s) on the B-party’s line receive a SETUP message, indicating the call type, the allocated B-channel and perhaps other information such as the called and calling numbers. If the call type is acceptable (in this case, “speech” or “3.1kHz audio”, A-law encoded and so on) and the addressing information is acceptable (e.g. the called number, sometimes called “MSN” or “DDI”, matches any number programmed into the phone by the user), the phones start to ring and each returns an ALERTING layer 3 message to the network, which passes it through to the A-party. The network normally also generates “ringback” tones down the B channel back to the A-party so he can “hear” the phone ringing.
A call has now been “initiated” all the way through the network, but not yet connected or charged for.
When the remote (B-party) user picks up one of the ringing phones to answer the call, that phone sends a CONNECT message to the network. The network returns a CONNECT ACKNOWLEDGE message to that phone (and RELEASE messages to any other ringing phones on that line to stop them ringing), and forwards the CONNECT message through the network to the A-party caller. The B channel is “cut-through” (connected) all the way from the caller to the called party and back, and the call charging starts.
The call is now fully connected.
When either of the users replaces the handset to end the call, that telephone sends a layer 3 DISCONNECT message to the network, containing a “cause” code, in this case cause 16, “normal user clearing”. See Q.850 or ETS 300 102-1 for a list of cause codes. The network disconnects the B-channel and returns a RELEASE message, indicating that all resources associated with the call should be released. The telephone responds with a RELEASE COMPLETE message when this is done.
Meanwhile, the DISCONNECT message has been forwarded through the network to the other user. The user may remain connected to the B-channel for a while to listen to any tones or announcements. When the user replaces the handset (or after a timeout), the telephone disconnects from the B-channel and returns a RELEASE message to the network. The network releases all resources associated with the call and returns a RELEASE COMPLETE message.
The call is now completely cleared.
Basic rate terminals can normally be plugged straight into an NTU (or direct to the line in North America) using normal “category 5”, “unshielded twisted pair” (UTP) network cable with standard “RJ45” plugs and sockets wired “straight through”. Note that the pairing of pins is important and not intuitive: the four twisted pairs of wires should connect to pin pairs 1-2, 3-6, 4-5 and 7-8. Polarity is important if more than one terminal is connected. Network cables must be kept away from mains cables.
If all terminals are less than about 10 meters from the NTU no further wiring configuration is needed. On a basic rate line configured as “point-to-multipoint” at the network, up to eight terminals can simply be connected in parallel at the NTU, otherwise only one terminal is allowed on “point-to-point” lines.
If a basic rate terminal is to be connected further than about 10 meters from NTU, 100 ohm termination resistors must be connected between pins 3 and 6 and between pins 4 and 5 at the far end of the cable, and the termination resistors inside the NTU disconnected. Companies such as RS sell suitable plugs with termination resistors and on the NTU there should be a switch that can be set to “OUT”. Terminals can be connected anywhere along this “passive bus”, but any “stubs” should be as short as possible (10m maximum).
Terminals can be connected further than 150m from the NTU, up to a maximum of about 800m, but they must be clustered together near the end of the bus and at the extreme length only one terminal can be connected. Termination resistors must be used at the end of the cable as described above and there should be a switch on the NTU that is set to “LONG” to enable “adaptive” timing in this configuration.
In Europe, terminals should be configured for “Euro ISDN” or “ETSI” or “CTR3” operation. In North America, a switch type and SPIDs need to be configured as described above. Other countries are often compatible with Euro ISDN, or may need a specific configuration (e.g. INS64 for Japan).
If more than one number is allocated to the line, you may need to configure your terminal to only accept calls to a specific number, sometimes called an MSN (multiple subscriber numbering), otherwise it will ring for all calls. Usually only the last few digits are necessary.
If outgoing voice calls are being rejected, try changing the configured call type from “speech” to “3.1kHz Audio” or vice versa. “Speech” calls are sometimes cheaper, but may not work for modem or fax connections.
Primary rate lines are also often connected using standard network cable and “RJ45” plugs, but using different pins. They are always point-to-point and therefore there is no need to worry about termination resistors or polarity. Older lines may connect using coaxial cable as described above, or with simple screw terminals.
Transmit and receive pairs are frequently presented the wrong way round, so some experimentation with “crossover” cables may be necessary. Primary rate crossover cables are not the same as basic rate or ethernet crossover cables!
Older primary rate lines sometimes do not support the standard framing scheme known as “CRC4 multiframing”. It may be necessary to disable this option in the terminal. There are also legacy lines such as “DASS2” lines in the UK that are electrically compatible and functionally similar, but completely different at layers 2 and 3.
A “loss of framing” (LOF) or “no signal” indication probably means there is no connection, or that the transmit and receive pairs are swapped. A “remote alarm indication” (RAI) means that the terminal is receiving a correct signal from the network but the network has detected a problem in the signal from the terminal. An “alarm indication signal” (AIS) means there is a signal from the network but the line is out of service.
The International Telecommunications Union (ITU) has all the international ISDN specifications. Those beginning with an “I.” or a “Q.” are particularly relevant. Country-specific specifications are often just a list of differences from these “base” specifications. Up to three can be downloaded free of charge.
The European Telecommunications Standards Institute (ETSI) has all the European ISDN specifications. They can be downloaded for personal use for free.
The Alliance for Telecommunications Industry Solutions (ATIS) has copies of many North American ISDN specifications. One copy of each can be downloaded for free.
Peter Ilieve’s uk.telecom mini-FAQ has many links to UK-specific information, including all the operators.
Ralph Becker has a comprehensive tutorial that covers similar ground to this one.
Dan Kegel’s ISDN Page has many useful links to other sites as well as a tutorial. Last updated in 1996, though.
Dialogic have a complete “Introduction to ISDN” course online.
Thomas A. Fine has another good tutorial.
Protocols.com is more technical, with a good list of country variants.
SEG communications has a practical UK-specific ISDN overview.