Norm Al Dude and Professor N. Erd
on the subject of ISDN

I have a hot topic for you: ISDN. But before I tell you about it, I have to show you some of the things without which no understanding of ISDN can be achieved.

Folks, Nicky Erd and I had a chat on how telephones carry voice. It happened because I was curious. When I sing in the bathroom under the shower, I swear my voice sounds like Pavarotti's. Yet, when I sang through the phone to my sweetheart, she said to me, "Stop oinking like a pig!" Could it be that the phone line was bad?

N. Erd explained that sound is a random analog signal which can be represented as a series of sine waves. The way I understood it is that my voice box is emitting lots of musical notes together, each of them being a "sine wave." A sine wave has three characteristics: amplitude (its value at each moment), frequency (of repetition), and phase (delay from some reference in time). Frequency is measured in Hertz (Hz) and is the number of repetitions (cycles) per second of the signal. What we humans can hear is called sound, and is made of sine waves between about 20 Hz and 20 KHz.

Figure 1: Phone line spectrum

The phone translates human voice into an electrical signal that only carries between about 300 Hz to 4000 Hz (it's all electronic magic), because it's cheaper this way and anyhow to recognize who's talking and the meaning of the spoken words, that's plenty of "bandwidth", as N. Erd says. Just to illustrate what I'm saying -- did you ever listen to a talk-show radio station? Come on! Admit it! You all listen some time to Howard Stern! Didn't Howard sound better than any caller? You bet! He used the whole 10 KHz bandwidth of the radio station, while the callers were using only 4 KHz. An analog electrical signal has a very complex shape and is very hard to transport through the network without distortion.

Back in the late twenties, a fellow whose name was Harry Nyquist thought of a way to take an analog signal (such as voice) and code it (just like with the Morse code) using ones (1) and zeros (0). For this, he invented something called a "CODEC" or coder-decoder. This thing that today is the size of a fingernail (a microchip) measures the input analog signal, codes the result of the measurement and sends this code down the telephpne lines and trunks. It does so often enough so its peer at the other end of the line can reconstruct the voice signal almost as good as it was at the calling side. N. Erd calls the measuring of the signal "sampling." Good old Harry Nyquist also recommended that the number of samples per second for a good representation of the signal has to be twice as big as the number of Hertz of the fastest sine wave contained in the analog signal. Since the telephone only allows 4 KHz through the phone line, sampling for voice is done 8000 times per second.

Each sample is coded in an 8-bit word, so 64 Kbps are needed for a digital phone conversation. Digital voice lines have a tremendous advantage over analog ones -- and here is why: as we all know from our high school science, electrical current that travels through wire attenuates. Therefore, we need amplifiers (every three miles or so) to boost the signal. Not only are these amplifiers expensive, but they are very dumb, too! They amplify both the voice signal and the noise (electrical interference) on the line! So -- if I'm looking at a three-mile line segment, I see here the signal, the new noise (picked up on this segment) and the old noise amplified (the noise that has been picked up on other segments).

With digital transmission, the line has no amplifiers. Instead, every mile or so, they put a repeater on the line. A repeater senses the incoming pulses that represent the "ones" of the code and repeats them over the next mile, without propagating the noise.

Figure 2: Why digital?

Back in the mid-fifties, Bell/AT&T who owned all telephone lines in this country of ours had another big problem: too much wire. Each new subscriber meant a new wire loop. And, to maintain a decent grade of service, the phone company had to add a new trunk for every so many subscribers. So, around 1956, Bell/AT&T invented their first 'digital transmission carrier' called T1, which made their wires work 24 times harder than wires used to work before T1.

T1 is a system that allows 24 digital telephone conversations to share (multiplex) the same 4-wire trunk circuit without interference (without somebody listening to your conversation). The 'bandwidth' of a T1 is 1,544,000 Mbps, with room for 24 x 64 Kbps for voice calls (1,536,000 Mbps) and the rest of the 8 Kbps to be used for keeping the ends of the trunk 'synchronized' -- a process called framing.

The big advantage of the T1 carrier system is that it can carry both data and non-data traffic, such as voice and video, because all these types of information are coded in ones (1) and zeroes (0). therefore, the T1 pipe can carry mixed 'integrated' traffic and a customer does not need to have separate expensive lines or trunks for voice, video, or data.

T1 lines can be leased to connect two customer sites permanently. This is OK as long as the customer has enough traffic to pass through the T1 pipe (and justify its cost). However, if no data, voice or video is transmitted (for example, at night) the customer still pays the monthly rent for the line.

It would be better if, instead, the customer could dial a number (just like with the telephone), and the carrier company would then connect the two sites on demand through their mesh of wires. This way, the customer transmits what is to be transmitted, then hangs up and gets a bill just for the duration of the call. This would be something that carrier companies call SDN or Switched Digital Network.

Before getting into ISDN, I have to explain to you how this business of T1 multiplexing works.

Remember, there are 24 digital sources of bits each of them pumping out 64,000 bits each second on a cable. Actually, there are two cables (four wires -- two transmit and two receive) that allow transmission by the two customer sites in both directions (this is what N. Erd calls full duplex).

At each end of the T1 line, there is a 'multiplexer' -- a sort of a funnel that allows the 2 sources to mix their bits together at the transmitting end and to extract their respective bits at the receiving end. What we have with T1 is a so-called 'time division' multiplexer. N. Erd says that each digital source is allocated a 'time slot'. During this 'time slot' (as small as about 125 millionth of a second) the source to whom this tiny speck of time is allocated sends 8 bits coded into pulses on the line (enough for a CODEC to send a voice 'sample', or for a data source to send a 'byte').

Figure 3: T1 multiplexing

The multiplexer (which may be called 'channel bank') mechanically stops at each source and collects the eight bits to ship them over the T1.

Source 1 sends eight bits, then source 2 sends eight bits, etc., etc., up to source 24. Altogether, 8 x 24 = 192 bits are sent. Right before the whole 192 bits, the multiplexer sends another bit called a 'framing bit' which is like a flag announcing 'here comes the train of bits.'

After that, the multiplexer starts all over again: flag, eight bits from source 1, etc., etc.

This train of 193 bits (flag and data or voice, or video) is called a 'frame'. Since the CODEC has to send 8000 samples each second (8000 groups of eight bits), the whole frame (for 24 sources) must repeat 8000 times per second.

If you take your calculators and multiply 193 x 8000, you get to the T1 rate of 1,544,000 bits per second. One of these sources that enters 64,000 bits every second is known as producing a DS0 or 'Digital Signal 0'. Combining 24 DS0s onto a T1 line results in a DS1 (Digital Signal 1).

Now we are almost ready for ISDN. Before that, however, I still have to explain something called 'signaling' and how it's done in most cases in today's telephone network in North America. "Signaling," N. Erd explains, "is the process of requesting services from the network; in addition, supervision is the process of reporting to a management unit the current status of the call." The way I see it, signaling is when I dial Grandma's number on the phone and supervision is some sort of a little devil that tells the telephone company once in a while "keep billing -- this conversation is still going." N. Erd says I have a simplistic view, but generally he agrees with me considering my feeble mind.

Well, folks, in today's digital network, telephone switches that make line connections, are hooked up to intelligent computers (called STP or Signaling Transfer Point Computers) that can order the switches to connect or disconnect lines and trunks.

As I pick up the phone to dial Grandma, the switch in the nearest telco office gives me dial tone to let me know I can send the number, and then, this nearest switch builds a data message that goes to the STP computer. This STP fellow looks at a map of the network and right away sees if there is a path or there is not path between me and Grandma. If there is -- then the same STP sends orders to all other STPs and switches needed to build the connection, and rings Grandma.

Figure 4: Common channel signaling system

If there is no path, or if she is talking, I get a busy tone right away. This kind of separate network of STP computers is known as a 'Signaling System' or a 'Common Channel Signaling System.' Various N. Erds wrote different programs to handle the calls. The current system is the seventh one and it's called 'SS#7', or 'Signaling System #7.'

Finally, we are ready to learn what kind of fish ISDN is.

N. Erd says that ISDN stands for Integrated Services Digital Network, and not for It Still Does Nothing or I Smell Dollars Now, as some might like to think. 'Integrated Services' means that the phone system supports a whole lot more than just you and me talking on the phone.

For example, when Grandpa got his triple bypass operation and was released from the hospital, he had his phone system tied to a medical emergency service. If he didn't feel well, he merely had to push a button and, as the phone rang at the emergency squad dispatch, it also rang his doctor and his number and address was displayed on a monitor as his medical record was printed on the screens of the two computers - at the squad and in the doctor's office, complete with his address and directions to his house.

Or, when the phone rings in the technical support room of a large computer firm, the screen on the operator's desk is already filling with all the data about the caller stored in a huge service data base. This way, the operator can greet the customer by name, and can intelligently discuss pending problems and escalation steps related to this account.

Many people hate to travel hours everyday to get to work, but like working for their employers. Today they can 'telecommute' -- in other words, they can have their cake and eat it too! By working at home, they save commuting time and aggravation and become more productive! And they can access all computer resources located at the company headquarters as if they were there themselves. While keying away at computer data, they can also talk on the phone with their customers or sweethearts, using the same line over which the data session takes place!

At home, Nicky's air conditioner is controlled remotely by the power company: At noon every day during the summer, they stop it for 2 hours, while monitoring the temperature in the house. If it raises by more than 2 degrees, they start it again. This way, they save on energy, and he gets a break on his electric bill of $16 every month! And they do it over the same phone that he uses to call his Grandma, and to work on his computer to surf the Internet.

It used to be that people had to travel sometimes thousands of miles to meet others, like for a sales meeting or for a seminar. Nowadays, a video camera and a large monitor are placed in each room where speakers or audiences reside and they enjoy an interactive session as if they were in the same room.

All these and some other wondrous accomplishments are possible because of this fish called ISDN.

It's all rather simple: ISDN is a network based on the T1 digital carrier with common channel signaling achieved through the SS#7 data network controlling the Telco switches to create digital paths on demand, to carry intermixed voice, data and video traffic. Brrr! I started to speak like N. Erd!

A user subscribes to two kinds of access: Basic Rate Interface for the home or small office; and Primary Rate Interface for the big headquarters.

They are both time-division multiplexed just like T1.

Figure 5: BRI and PRI pipes

The Basic Rate Interface or BRI, however, instead of 24 DS0s, accommodates only 2 (each of 64 Kbps) called B or Bearer channels, plus another channel called D (Delta) of 16 Kbps to be used for signaling. That's why the BRI is also known as a 2B+D interface to ISDN.

The Primary Rate Interface or PRI is also known as 23B+D where B is a 64 Kbps Bearer and D is this time also a 64 Kbps channel to be used for signaling. (That's only in North America; Europeans have their own system with 30B+D -- they always play smart!)

If you recall, in T1 we had 8 bit times allocated to source 1, 8 bit times to source 2, etc., up to source 24 that was also allocated 8 bit times to transmit. In addition a framing bit was sent. And the whole thing was repeated over and over. PRI works exactly like this, except for the last source which is now special: it's smart enough to do the signaling (request for services, dial phone numbers, report status) on behalf of all the other 23 sources.

BRI also works this way, but since only 2 Bs of 64 Kbps and 1 D of 16 Kbps are present, a frame here allows 8 bit times for B1, 1 bit time for D, 8 bit times for B2, 1 bit time for D, plus framing and other control information.

The wise people from various standard committees like CCITT, now known as the TSS (Telecommunications Standards Sector) of ITU (International Telecommunications Union) have described the BRI ISDN access as follows:

Figure 6: CCITT's ISDN architecture

A two wire bus called 'U' (only in North America) up to 18,000 ft long, with a minimum wire size of 26 AWG, links the telco switch with the customer premise. The box into which the U bus terminates is called NT1 (Network Termination 1) and you can think of it as the 'ISDN BRI modem'.

From the NT1, a four wire full duplex bus called T goes into another box called NT2 (Network Termination 2). The NT2 can be your PBX (Private Branch Exchange - your own phone or data switch), a bridge/router, a multiplexer, etc. Out of NT2 another 4 wire bus called S connects to the ISDN stations or terminals.

There are two types of such stations. ISDN-capable ones are called TE1 (Terminal Equipment 1). The other type includes things that you and I use daily, such as PCs, dumb terminals and plain old telephones that cannot signal over the D channel and cannot transmit at a rate of 64 Kbps. Those are called TE2 (Terminal Equipment 2). To hook up to the S bus, they need a special standalone box or a card inserted in the PC, called a TA (Terminal Adapter). The interface between the TE2 and TA is known as the R interface.

I like to think of U as the 'User' interface, T as the 'Transmission' interface, S as the 'Station' interface and R as the 'Rate Adaptation' interface. Sometimes, when stations hook up directly to the NT1, there is no NT2, and the S and T busses are combined into one called S/T.

Up to eight TEs maybe connected to the same NT1 and BRI in what they call a 'passive bus arrangement' in a multipoint configuration. Only two of the eight devices can talk over the two B channels at a time; the others have to wait patiently for their turn.

For the PRI, the access is regular T1 -- through a CSU (Channel Service Unit) instead of an NT1. After this, everything is like in the BRI case - NT2, etc.

Each BRI or PRI can get one or more phone numbers; if one is supplied, then the telco switch treats the B channels as part of what they call a 'hunt group' and the incoming calls are allocated the first available B channel. Each attached device is also allocated an extension called a TEI or 'Terminal Equipment Identifier'.

Figure 7: BRI multipoint passive bus

At power-on time, the switch and the TE/TAs exchange the necessary information to bring to 'active state' all the 'service profile features' associated with that subscriber device. In some switches this is done automatically, based on the phone number; in some other, it is the TE that has to report to the switch its 'SPID' or Service Profile Identifier'. Once the switch verifies and validates the SPID or caller ID, the line is up and ready for use. Periodically, the switch and the TEs exchange short messages to check the line status. All this activity is done via the D channel.

When a TE wishes to place a call, it does so over the D channel by sending a message named 'Call Request' in a language very similar with X.25 (says N. Erd). Then, the switch in the Telco forwards the message to its attached STP computer which uses SS#7 to establish the connection. The called side is presented with an Incoming Call Indication packet. Then, if accepted, the calling device receives a Connected Indication packet, and it can start transmission over the allocated B channel. Otherwise, the call will be rejected and a code with the reason for the rejection will be provided to the calling TE by the switch.

ISDN tariffs are written so that one can signal the network via the D channel to request more bandwidth for a connection. When available, the network assigns additional B channels to the call. The connecting equipment combines the B channels to create H (or Hyper) channels. There are many possibilities, the most common ones being 384 Kbps, 768 Kbps and 1,536 Mbps. This way, full motion video can be accommodated over several such channels combined into a larger bandwidth pipe. This is called 'inverse multiplexing'.

Also, if one has a terminal adapter that can 'speak' packet X.25, it can do so via the D channel while no signaling is in progress, and while using the circuit-switched B channels for other calls (examine stock exchange data via CompuServe, while calling Grandma and reading a file from the office). And, at the same time, the power company can control the meters and the usage of electricity over the same D channel through a process called telemetry. Or your local security company can monitor your security system over the same D channel.

But, perhaps the hottest application today is telecommuting. Just imagine yourself working from home at some very large CAD/CAM project (you may be an architect or mechanical designer). Large files need to be downloaded from remote servers to your workstation. With an ISDN bridge, your workstation is connected to a BRI line for anywhere between $20 and $60/month (and in many places there is no measured local service charge) depending on your local tariff. Long distance charges are whatever you pay today to your long distance carrier.

At the headquarters where the server is located, there is a really powerful bridge that can receive calls from hundreds of telecommuters like you and can concentrate them for broadcasting over the central LAN. As you logon to your server, the bridge interprets the logon message and dials over the D channel to make the connection over B1. The file transfer begins, and as data accumulates in transmit and receive buffers and exceeds certain programmable thresholds, a second call is placed over the B2 channel. B1 and B2 are now used as a thicker 128 Kbps pipe.

With 8:1 compression, the bridge effectively acts as a 1 Mbps transmission/reception device! The file comes to your workstation nearly as fast as if you were directly attached to the server.

Meanwhile, the bridge also has a phone plug. If you wish, you can call Grandma. As you pick up the phone and dial, the D channel carries Grandma's number and, as the connection is made, the file transfer is left with 64 Kbps over B1 only, while you talk via B2. As one of you hangs up, if the data transfer is still running furiously, the B2 channel is reallocated to the file transfer (128 Kbps again). If somebody else calls you and you bought the line with 'call waiting', your phone is rung via the D channel and, if you pick it up, B2 is allocated to the voice call, while the file still goes via B1, etc.

By the way, with Caller ID (which ISDN and SS#7 made possible), your TE (bridge) may accept or reject calls as programmed by you! And if no activity is seen on the ISDN line for a programmable period, all calls are dropped to save you money. Pretty neat, huh? N. Erd has one of these gizmos from a company called Gandalf and is extremely happy! He says "the investment was paid off in less than a month, and the amount of time it saves him is incredible."

The truth is ISDN is not new. It all started 20 years ago. But now, for the first time, carriers and vendors alike started to market it properly -- and lines are available almost everywhere (wherever ISDN service is not offered, it is possible to interconnect non-ISDN lines such as analog modem-based or switched-56 service to ISDN.)

It all makes sense and vendors, carriers and, most important, users alike can profitably use it! It seems to be the choice technology today for normal data, voice and video-conferencing integration.

You're welcome to peruse a list of publications that offer more information on ISDN.

So long for now,
Norm Al