by Mike Starkenburg
T-1, fractional T-1, and T-3 services are high bandwidth digital data transmission systems originally designed for carrying voice calls between telephone company central offices. Today, the T-carrier system is used for transfer of voice, data, and video signals in business applications of all sizes.
T-1 lines can connect distributed offices or a corporate office to the Internet, at speeds ranging from 1.5Mbps up to 44Mbps.
When the telephone system was originally designed, it was built on the premise that a large percentage of calls would be handled locally. Within several thousand feet of every subscriber, there were central offices that handled the connections of calls. Often, the originating party and receiving party were both serviced by the same central office, so the call could be switched locally.
If the two parties were not served by the same central office, the phone company handled the connection between two of their offices. These connections between offices were called inter-office trunks (IOT).
Over time, call patterns became more diverse; people were using the system to make calls farther and farther away. As the number of long distance calls went up, it increased the demand for IOT connections.
The original IOT connections were made over an analog system called the N-carrier. The N-carrier system used an analog device that multiplexed calls, allowing 12 voice calls to share the same physical wire. Unfortunately, the system quickly outgrew this capacity. To avoid running more expensive wire between central offices, the phone company looked for ways to increase the number of calls it could squeeze into existing wire.
The L-carrier system, which was used in the late 1960s, had a much higher capacity. Using coaxial cable, Bell was able to carry up to 6,000 calls on one wire pair. To run coaxial cable long distances, however, the phone companies had to place amplifiers periodically along the line. These amplifiers were expensive, unreliable, and produced poor quality calls. In addition, running coaxial cable was more difficult and expensive than running the simple copper wire used by the N-carrier. In the early 1970s, the phone companies began using digital systems to solve these problems and increase capacity even further.
The T-carrier system was the first widely deployed digital transmission system. The phone companies placed hardware in the central offices that converted the analog voice signal to a digital bit stream. This bit stream could easily be multiplexed onto a single carrier. Using the T-1 version of the T-carrier, 24 calls could be carried on two pairs of inexpensive copper wire. The quality of the digital signal did not degrade over long distances, and the copper wire could use simpler amplification hardware. Over time, the phone companies deployed higher capacity versions of the T-carrier system, such as T-3, to meet increasing demand.
Originally, the T-carrier was used only as an IOT. End-user services were still available only in analog format. In the mid 1970s, the phone companies reluctantly began rolling out T-1 service to large businesses. The primary use of this was for businesses with large call centers and 800 numbers. Digital service was expensive and time-consuming to install, because the line needed to be specially conditioned by the phone company.
NOTE: Another type of digital line was made available to the public before the T-carrier system. Digital Data Service (DDS) was available for data transfer of up to 56Kbps. Because a T-1 was only 4 to 8 times more expensive than DDS but carried over 24 times the bandwidth, it quickly became the data service of choice for businesses.
Today, T-carrier service is affordable and widely available. The rise of the Internet and the growth of e-mail as a communications medium has made the T-1 a household word.
The T-carrier system is a bipolar, framed format, full-duplex, channel-based digital communications system. In this section, we'll explain how those technologies work together to produce a reliable digital connection.
A digital signal is made up of ones and zeros. In the T-carrier system, these ones and zeros are represented by changes in voltage along the wire. In the simplest implementation of this, you can imagine that there would be a positive voltage for every one, and no voltage at all for every zero. The problem with this kind of encoding (called unipolar) is that there is no "return to zero" between positive pulses. Without some sort of "return to zero," the signal can lose track of how many positive "one" pulses occurred, called loss of synchronization. Figure 16.1 shows an example of a unipolar signal.
FIGURE 16.1. An example of unipolar line coding.
To counter this, the T-carrier system uses a process called alternate mark inversion (AMI) or bipolar encoding. As Figure 16.2 shows, bipolar encoding still uses no voltage to represent a digital zero, but uses alternative positive and negative voltages to represent digital ones. This system adds redundancy to the timing of the circuit.
FIGURE 16.2. An example of bipolar line coding.
The only remaining danger with the AMI system is that a large number of sequential zeros will cause a loss of synchronization. The ones density rule states that no more that 15 zeros could be transmitted consecutively. Obviously, in a random bit stream, it is easy to imagine that there would be more that 15 zeros in a row.
Originally, the phone companies used a technique called pulse stuffing where every eighth bit was taken out of the signal and forced to be a pulse. This reduced the bandwidth of the channel from 64Kbps to 56Kbps.
To get back the use of that additional bandwidth, the pulse stuffing method of synchronization was replaced by the bipolar 8 zero substitution (B8ZS) method. Using this method, the hardware at either end of the signal listens for eight consecutive zeros in the signal. When it finds them, it replaces the entire 8-bit word with a fictional word.
NOTE: A T-3, because it is so much faster, needs more frequent synchronization than a T-1. The B8ZS system is far too slow. A T-3 uses a B3ZS system, which works the same as B8ZS but modifies any group of three zeros instead of eight.
To differentiate between a real word and a fictional one, the hardware creates a bipolar violation. In the AMI system, every consecutive "one" pulse alternates between positive and negative voltage. By sending two consecutive positive pulses, the system signals that there is a fictional word coming. Figure 16.3 shows an example of a bipolar violation.
FIGURE 16.3. An example of bipolar violation.
To assist in synchronizing the devices on the T-carrier line, the phone companies developed a framing system. The early frame standards--called D1, D2, and D3--were used only for analog data and voice over a T-1. Today, there are two primary framing standards for the T-1 circuit, called D4 and extended superframe (ESF). The T-3 circuit uses another protocol called M13.
In Figure 16.4, you can see an example of the D4 framing format. One framing bit prefixes 24 bytes of data, each from one of the 24 channels in a T-1. The framing bit follows a special 12-bit pattern, called the frame alignment signal. Every 12 frames, the signal repeats, allowing the hardware on either side of the connection to signal changes in line status. This group of 12 D4 frames is called a superframe.
NOTE: Why is a T-1 a 1.544Mbps connection if 24 64Kbps channels totals 1.536Mbps? D4 framing adds one bit for every 192 data bits, so it adds 8Kbps to the total T-1 line.
FIGURE 16.4. The D-4 Framing Format.
In the early 1980s, AT&T decided that it could make better use of the 8Kbps frame overhead. By extending the superframe from 12 frames to 24, AT&T found that it could provide sufficient synchronization in only six of the 24 framing bits. Six of the bits were used for error correction, and the remaining 12 were used for network monitoring. Figure 16.5 shows the extended superframe format.
A T-3 uses a framing system called M13. Because the T-3 is much faster than a T-1, its frames are much larger. The M13 frame uses a 4,760-bit frame, compared to a 193-bit D4 frame. Of this, 56 bits are used for frame alignment, error correction, and network monitoring.
FIGURE 16.5. The ESF Frame.
As the T-1 became available commercially, a large number of companies began producing hardware to maximize the benefits of the carrier system. Because the T-1 is made up of 24 separate channels, configuration of the system is very flexible. Voice, data, video, and other signals can all share the same T-carrier. Allowing all these signals to use the same transmission carrier is called multiplexing.
A T-1 multiplexer can simply manage the data input sources, or it can combine channels for high bandwidth applications. Today, for example, it is common for the entire T-1 to be used for a single 1.536Mbps data application.
Figure 16.6 shows a typical multiplexer installation with a T-1. Note that the analog-to-digital conversion of the voice lines can occur in either the PBX or the multiplexer, depending on the exact hardware installed.
FIGURE 16.6. A typical multiplexer application.
The T-carrier system uses time-division multiplexing (TDM). As the name implies, TDM uses units of time to divide the signals from each source. These units of time, or timeslots, are approximately 1/8000th of a second, which corresponds to the sampling rate of an analog voice call.
On a T-1, there are 24 carrier channels, each transmitting 8,000 timeslots/sec. The multiplexer is responsible for interleaving these timeslots so that the receiving device will be able to properly route the contents of each timeslot.
If four devices are all sending data through a TDM, the data is combined sequentially and sent down the T-carrier. At the receiving end, the order tells the de-multiplexer which device should receive the data. If a specific device has no data to send in a specific timeslot, then the multiplexer adds a null carrier to keep the ordering consistent.
Because TDM allows some timeslots to go unused, it is not as efficient as it might be. A newer multiplexing technology, called Statistical Time Division Multiplexing (STDM), is being developed to fix this inefficiency.
To carry a voice signal on a digital carrier, the voice signal needs to be converted to digital pulses. First, the signal is sampled, or measured many times per second. That measurement is then quantized, or converted into a number. The number is rounded to the nearest number that can be represented in an 8-bit byte. This process is called Pulse Code Modulation (PCM).
While the human voice can produce sounds in the range between 50 and 1,500Hz, most of what we hear is between 300 and 3,400Hz. This 3,000Hz of bandwidth, plus 1,000Hz for separation between calls, gives a total of 4,000Hz.
The rate of sampling is determined by a rule called the Nyquist theorem. The Nyquist theorem states that, to ensure accuracy, a signal should be sampled at twice the rate of its frequency. In the case of voice signal, the proper sampling rate is 8K/sec.
Because each sample is 8 bits and samples are taken 8,000 times per second, the total bandwidth necessary to carry a digital voice signal is 64Kbps. This PCM requirement is the building block for both the DS0 channel size (described in "The Digital Signal Heirarchy" section of this chapter) and the ISDN B-channel size (described in Chapter 13, "ISDN").
A newer PCM standard promises to increase capacity even further by halving the amount of bandwidth needed to carry a voice signal. Adaptive PCM (ADPCM) predicts the next voice sample based on the level of the previous sample. The limitation of the technology is the inability to reliably transport analog modem traffic faster than 4,800bps.
A fractional T-1 is nothing more than a standard T-1 in which only some of the 24 DS0 circuits are available to the end user. When a fractional T-1 is ordered, the carrier installs a full T-1 to the location, but provisions only the requested number of 64Kbps channels. The remaining channels are not configured. A fractional T-1 uses the same terminating hardware as a full T-1.
Fractional T-1s are useful when fewer than 1.544Mbps of bandwidth are needed, because they are less expensive than a full T-1. They are also useful when bandwidth needs are expected to grow. Because a full T-1 and the appropriate hardware are already in place, increasing a fractional T-1 requires only a phone call to your carrier.
In the world of digital communications, there have been two kinds of standards. AT&T, by providing much of the research into the area, has set de facto standards that other manufacturers have picked up. The other standards were more formalized, and were released periodically by associations and standards bodies. Two high-level standards which particularly affect T-carrier lines include the Digital Signal Hierarchy (DSH) and the OSI Network Model.
The different levels of bandwidth offered by the T-carrier system were standardized by the American National Standards Institute (ANSI) in the early 1980s. Table 16.1 shows the DSH.
| DS Level | North American Bandwidth | Voice Channels |
| DS0 | 64Kbps | 1 |
| DS1 | 1.544Mbps | 24 |
| DS2 | 6.312Mbps | 96 |
| DS3 | 44.736Mbps | 672 |
| DS4 | 274.176Mbps | 4032 |
The DSH is based on a core channel called the DS0 (digital signal 0). The DS0 is usually a 64Kbps channel, which is the amount of bandwidth needed to carry a single digitized voice signal. DS0 signals are aggregated to provide higher levels of bandwidth. Common DS0 levels include
In Europe, the DS0 standards were set by the International Telecommunication Union, and are therefore slightly different. While the core DS0 channel is the same size, the various levels are aggregated differently. As an example, the European equivalent of the T-1 circuit, called the E1, carries 30 DS0 channels instead of 24.
NOTE: You may hear people in the industry using the terms T-1 and DS1 interchangeably. Technically, the T-1 is the North American implementation of the DS1 standard. You're going to get your point across either way. In Europe, however, the DS1 service is called E1.
The T-carrier system addresses the first three layers (physical, data link, and network) of the OSI model. The higher layers are addressed by the hardware and software in the user's network.
The physical layer of the OSI model defines the hardware, wiring, connectors, and other electrical and physical characteristics of a network connection. In a T-carrier connection, this includes the cabling and Channel Service Unit/Data Service Unit (CSU/DSU) hardware. The CSU/DSU handles the physical and electrical termination of the connection and monitors physical line status.
The OSI data link layer handles connection maintenance and error correction. A T-carrier framing system (either D4, ESF, or M13) handles the data link functions.
In the OSI model, the network layer defines signaling procedures between the network and users. In a T-carrier connection, the framing system provides the basic signaling pathway, and the CSU/DSU handles signal generation and management.
The hardware involved in a digital leased line has become progressively simpler as service became widespread. Today, the major hardware categories in a customer installation include cabling, service units, multiplexers and channel banks, and data terminal equipment.
In an ideal installation, a T-1 operates over special low-capacitance shielded twisted-pair cabling. In many installations today, however, the phone company tries to use existing unshielded twisted pair. In these cases, extensive testing is required to ensure that the line is clean enough for T-1 use.
A typical T-1 line consists of two twisted pairs of copper wire. One pair handles data to the subscriber; the other pair handles data from the subscriber. These connections are usually terminated in an RJ48 jack.
Although analog signals over twisted pair can usually travel up to 18,000 feet from subscriber to central office without amplification, T-1s require cleaner lines. Repeaters are placed about every 6,000 feet to power the signal. The repeaters do not amplify the signal, but rather duplicate the digital signal to the next segment of the line.
The twisted-pair cable cannot, however, carry the increased bandwidth of the T-3 carrier. Both T-1 and T-3 carriers can be provided on alternate media, including coaxial, fiber, microwave, and infrared. Table 16.2 shows the approximate capacity of these other types of media.
| Media | T-Carrier Capacity |
| Twisted pair | Up to 1 T-1 circuit |
| Coaxial cable | Up to 4 T-1 circuits |
| Microwave | Up to 8 T-3 Circuits |
| Fiber optics | Up to 24 T-3 circuits |
Generally today, the functions of both the DSU and the CSU will be found in one device, commonly called the CSU/DSU. The CSU/DSU may be provided or recommended by your phone company. In some installations, you may find separate units, so we will describe the functions separately here.
The first piece of hardware in the customer premises is the CSU. The CSU serves a number of functions:
The primary purpose of the DSU is to convert the standard unipolar digital signal from the multiplexer into a bipolar signal. The DSU also controls timing and synchronization of the signal.
The multiplexer handles the combination and sequencing of several different devices into the signal. The earliest multiplexer was known as a channel bank. The channel bank had one hardware interface for each DS0 channel in the T-1 circuit. This fixed hardware configuration limited the channel bank; even if your application needed only 9.6Kbps, such as an older 9,600 baud modem, the channel bank would use an entire 64Kbps channel to pass this signal.
Today, a multiplexer can super-rate a channel for more than 64Kbps, or subrate a channel for less than 64Kbps. Multiplexers can take input from a variety of terminal equipment, and can be software-configured. A multiplexer may be able to provide internal PCM encoding for analog devices.
Depending on your application, you may want to multiplex a number of different sources. These devices are referred to collectively as terminal equipment, and serve as the user interface to the network. Your terminal equipment might be a network router, a video conferencing system, or a PBX.
The most common piece of data hardware connected to a T-carrier line is a network bridge or router. A network bridge connects two dissimilar networks, providing the necessary protocol translation between them. The most common type of network bridge connects an CSU/DSU to an Ethernet local area network (LAN).
Most T-1-to-Ethernet bridges provide some sort of routing ability. Routing is a more complex version of bridging, where each packet is analyzed to determine if its destination is on the LAN or across the T-1 line. Depending on your specific hardware choice, you can find T-1-to-Ethernet bridges that route several different protocols.
Which protocol your hardware supports depends on your installation. If you are linking two AppleTalk LANs in different locations, you'll need a router which supports this protocol. Similarly, if you want to connect two Novell networks, you'll need a router that supports the IPX protocol. To connect your LAN to the Internet, you need to route TCP/IP.
Routers generally provide one serial connection for the CSU/DSU, and an RJ45 connection for your Ethernet LAN. Some vendors, such as Cisco, sell routers that have integrated CSU/DSU hardware. These pieces of equipment are the simplest way to connect your LAN directly to a T-1.
T-carrier lines are often the best choice for data applications with constant demand for a high level of bandwidth. If your application meets these requirements, you next need to secure a leased line solution.
T-carrier connectivity is provided by the regional Bell operating companies for short hauls, and by national telephone companies such as Sprint and AT&T for long-distance connections.
FCC regulations require that you provide the carrier with a certain amount of information when ordering service, including:
In addition, the FCC may require an affidavit certifying that your terminal equipment is compatible with the network you will be using. Your terminal equipment manufacturer may provide a sample affidavit for you to complete.
Costs of T-carrier circuits vary by provider, but generally are charged based on the length and total bandwidth of the circuit. Local circuits have a monthly cost per mile for a fractional T-1 circuit, with each additional 64Kbps channel activated adding to the cost, up to full T-1 bandwidth. For long distance connections, charges are generally a flat rate per month, and are costed by city pair. For example, a T-1 from New York to Los Angeles is a higher flat rate than a T-1 from New York to Baltimore.
If your application requires a fairly constant amount of bandwidth, a leased line is generally the most cost-efficient connection. If your bandwidth needs vary widely during the day, you may want to consider an ISDN connection, which charges by the minute per 64K channel in use. This method of call billing allows you to only pay for a high bandwidth connection when it is needed, instead of paying the full flat rate.
A popular use of T-1 and T-3 lines are to connect a LAN or a server to the Internet. The T-1 line provides enough bandwidth for a large number of end users to send e-mail or browse the Web.
If you do not have any connectivity to the Internet, consider using your Internet service provider (ISP) as a "one-stop shop." Most ISPs who provide leased line services have good relationships with the local digital service providers and can expedite your installation. In addition, an ISP will be able to assist with administrative functions such as domain name registration, IP address assignment, and network monitoring. Most ISPs also offer some other services, such as newsgroups access and Web hosting.
Once you connect your LAN to the Internet via a digital leased line, the sensitive data on your computers becomes vulnerable to unauthorized access. Before getting "wired," investigate firewalls, router filters, and other methods of security. Your ISP should be able to assist you in securing your network, or can refer you to a network security expert.
TIP: The LIST features details on more than 5,000 ISPs.
T-1 lines are heavily tested upon installation, because there are so many places where failures can occur, especially in the local loop. Often, though, failures can occur after installation as well. Equipment failures, cable cuts, and electrical interference can cause total failure or distortion in the lines, even after months of excellent performance.
Many CSU/DSU devices include loopback testing functions that help diagnose these problems, but in general, there is little the end user can do to troubleshoot a T-1 line. If your connection is down, or you experience poor performance and connection failures, you should contact your carrier.
The explosive growth of the Internet has made high-bandwidth connectivity a widespread need. To answer this need, The T-carrier circuit has developed from a phone company proprietary trunk standard into a common WAN building block. Today, T-1s, fractional T-1s, and T-3s are widely available and affordable methods of interconnecting office LANs and connecting to the Internet.
Bandwidth, bandwidth, and more bandwidth. As new applications demand more capacity for data transmission, you can count on digital leased lines, including T-3s, to become even more common. But for applications requiring more than the 44Mbps of a T-3, several new technologies are becoming available.
The broadband ISDN standards, which are still being developed, aim to provide true digital bandwidth-on-demand. B-ISDN supplies bandwidth in excess of 1.544Mbps, or faster than a T-1. B-ISDN will be found in three common forms:
The other major standard that will affect digital service in the near future is the Synchronous Optical Network (SONET). SONET uses a new standard for framing that allows massive aggregation of DS0 channels. As shown in Table 16.3, SONET is designed to provide bandwidth capacity in the gigabit range. While the technology is currently used only by the telecommunications companies, you can imagine that the service will follow the example set by T-1 and T-3 lines, and eventually become a public service for large companies.
| SONET Level | Maximum Bandwidth |
| OC1 | 51.84Mbps |
| OC3 | 155.52Mbps |
| OC9 | 466.56Mbps |
| OC12 | 622.08Mbps |
© Copyright, Macmillan Computer Publishing. All rights reserved.
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}