This chapter discusses geographically very large networks. From the previous chapter, Fibre Channel is suitable for delivering data over distances of up to 10 km. ATM functionality can be mapped into fibre channel so that ATM is transported over fibre channel links. In still larger networks, ATM is used as a mode of communication between nodes, where a node is best described as a switch that manages multiple connections.

The history of ATM within the telecommunications companies indicates its position as a backbone technology within telecommunications infrastructures. Descriptions of ATM do not complete the telecommunications picture, however, and this chapter presents some of the missing components relevant to data communications.

8.1  Introduction

The early phone network consisted of an analogue signalling system connecting telephone users directly via interconnecting wires. The term plain old telephone system (POTS) was coined in the US to describe the service of voice communication through a telephone handset and sometimes the infrastructure supporting that service. One important aspect of this service is the ability to use the telephone during an electrical failure.

The POTS infrastructure is inefficient and prone to breakdown and noise because of the age and variation of environments in which it is installed. As a result, it does not lend itself easily to the transmission of data, especially over long-distance connections. Packet based digital switching systems began replacing the old infrastructure in the 1960s, but due to their size and complexity, the final connection from the local central office to the customer equipment is mainly an analogue connection.

8.1.1  Character

In classifying networks by geographical distance, and arriving at telecommunication networks, further categorization by the service that is offered upon them is useful. For instance, voice, data, message, or image transmission, are all telecommunications services. Networks in this chapter may also be local or global, international, national, or within a municipal area, in scope. Different physical layer techniques are used, such as cable, radio frequencies, and microwave, depending upon the installation and specific purpose of the link.

In terms of data transmission, these connections serve as the interface between LANs and generally include the main backbone of telecommunication networks. The relationship between the types of network are shown in Figure 8.1 and the pattern created by the links might be likened to the delivery systems for electricity or water, transport networks such as road and rail, or even the blood vessels in the body.

Figure 8.1 shows how a close-up view of how one section might look in the neighbourhood of Example Studio, and the pattern of increasing complexity continues as the data filter out to encompass the LAN and the individual computer that is generating or receiving the data. It is possible to see the pattern of increasing complexity towards the edge of the network as the detail level is increased. This phenomenon is called edge-centric, and can be thought of as being fractal in nature. Interesting research on the structure of networks can also be found within the Internet2 project (Internet2 International – see Notes and further reading). The fractal nature of networks has also been explored at British Telecom (Appleby, 1994).

8.1.2  The Line of Demarcation

At the highest level of detail, the mapped data are moving within a private network belonging to Example Studio. The point where responsibility for the transmission of data passes between the institution and the telecommunications provider is called the line of demarcation. The provision of services in this instance is normally documented within some agreement, which includes the service levels. Although the line of demarcation may be moved to unfamiliar locations, such as at the edge of the customers’ property, the dependency is upon the exact nature of service that has been purchased.

A telecommunications provider can only make guarantees whilst the data are travelling on their networks. If a connection is required between studios on different continents, the choice of service provider becomes important since different providers may use different mechanisms to transmit the data.

This chapter defines telecommunications networks as those parts of data transmission outside of the line of demarcation. In more detail, a telecommunications network is defined as interconnecting a number of stations using telecommunication facilities. The physical circuit between two points is referred to as a link, as opposed to interface, used so far. Confusingly, a node is no longer a single device but a point of junction similar in function to a hub.

8.1.3  Circuit Switching

When considering packet switching networks as discussed in Chapter 3, data are expected to be in some message format containing destination addresses. Messages are divided into packets and transmitted over a network of cables interconnected by decision-making devices, which direct the data to their destination.

8.2  Public and Private Telecommunications Networks

There are many telecommunications networks in operation, providing a wide variety of services. A list of these would certainly include public switched telephone network (PSTN), private-line voice networks (PVN), private-line data networks (PDN), packet switching networks, and public switched data networks (PSDN). Services available include audio programme networks such as cable radio, as well as video programme networks, such as cable television and video-on-demand.

The two categories of interest are public switched networks and private- or leased-line networks. The first of these classifications is variously known as public, switched, public switched, message toll service, long distance, direct distance dialling, and interexchange facilities. The following terms are used to describe the second of these main categories: private line, leased line, dedicated line, full-time circuit, and tie line.

8.2.1  Public Switched Networks

Public switched networks provide business and residential telephone services for voice and data transmission to the general public. Users share common switching equipment and channels, and callers wait their turn for service if all the facilities are in use. Fees are typically paid for the use of the network based on the time the call takes, and how much distance it covers, and is very much in the spirit of the traditional commercial telephone model.

Several approaches to the transmission of data over public networks have been taken, including the installation of packet switching public networks to provide a more efficient method of transferring data over networks, as well as the digitizing of voice data to send over private networks.

8.2.2  Private Networks

Private networks fall into two distinct groups. Those that are truly private, and those that are leased from, or at least managed by, a telecommunications provider for sole and exclusive use by the customer. A truly private network means responsibility for the entire 7-layer model associations including the transmission media such as cables. Due to the economics, truly private networks on the global scale are relatively rare.

Leased services are more readily available to enterprises needing general communication with a limited number of other locations, or with the Internet. With POTS-type arrangements, services are implemented within the telecommunications provider as a foreign exchange service (FX), implying a telephone exchange ‘foreign’ to the exchange area where the user is located. A private leased line connected to subscriber premises implements this service to a central office in another exchange. The subscriber can then make an unlimited number of calls to any number within the foreign exchange area for the cost of a local call.

Within data networks, the subscriber normally leases a single point to point connection, of a fixed QoS, calculated on a bandwidth per mile basis. Although these arrangements appear to offer point to point connection, the bandwidth is assigned within the exchange, and a permanent connection is opened through the switches.

8.2.3  Virtual Private Networks

One valid definition of virtual private networks (VPN) is the use of virtual IP subnets, where a workgroup of computer stations has the same IP subnet address, although data are transmitted on different physical network cables, passing data through a router if necessary.

In the context of telecommunications, a VPN is a mixture of public and private network circuits in a customized arrangement, implying some security for the connection. A leased line from a telecommunications company includes a bandwidth assignment made between named points and assigned at the exchange. Security can be applied using some encryption method.

Virtual private networks are favoured in franchise arrangements, where a cheap/fast general-purpose internetwork connection can be secured over public lines. The central office cannot be dialled into or accessed via the Internet (subject to the exact specification) and so the perception is that the link is free from hacking.

8.2.4  Managing the Service

From the point of view of an audio recording studio, as opposed to a broadcast service provider, the responsibility for data transmission once it is outside of their premises, belongs to the telecommunications companies, according to some QoS agreement. Arrangements can be made to lease or rent lines depending upon the national circumstances, but the connection over international boundaries will often be managed by different third parties along the route.

In order to transmit audio over the various connection packages available, it is necessary to understand what these services offer in terms of bandwidth, compatibility, and delay, the key indicators of the quality of the service. Comparison of the figures within the service agreement, against the figures generated by work which the link will be put to, will then give a realistic assessment of whether the proposed investment will meet the requirement.

8.3  ISDN

8.3.1  Introduction

Integrated services digital network (ISDN) is an international concept whose objective is a digital, public telephone network. The implementation is a digital data service capable of transporting any type of data using technology essentially similar to regular telephone calls. The architecture to support this objective builds upon the old POTS infrastructure, although the installation of compliant cables is sometimes required. ISDN is capable of offering digital services to a wide range of consumers and the link allows data to be transmitted using end-to-end digital connectivity.

The form of ISDN of most interest to audio professionals is basic rate interface (BRI), as being the cheapest and most realistic publicly available service. The primary rate interface is the same service anyway, but multiplied by 30 times or so. This is clarified later.

Subscribing

To access the ISDN service, it is necessary to enter a subscription arrangement to an ISDN line. The premises to which the service is to be installed must be within 5.5 km of the telephone company central office for the service to operate correctly. If the service is to be installed in premises outside of this limit, repeater devices are required, or the ISDN service may not be available at all. It should be noted that the distances vary depending upon the international variation in ISDN specifications.

Also required at the customers’ premises is the equipment to communicate with the telecommunications provider switch, and with other ISDN devices. These devices include ISDN terminal adapters (sometimes referred to as ISDN modems) and ISDN routers. For audio purposes, a coder–decoder (codec) is generally purchased which formats the incoming signal for ISDN.

There are some differences between ISDN standards in different countries, and this has been one cause of slow consumer uptake of the service.

Provision

There are two basic types of ISDN service: basic rate interface (BRI) and primary rate interface (PRI). BRI is supplied at the line of demarcation in the customer’s premises, in the same fashion as a traditional telephone point, but where an adapter must be installed, usually by the provider. It consists of three channels: two channels are used for actual voice or data traffic with each one operating at a rate of 64 KB/s. These are called bearer channels, or B-channels for short. The third channel is used for call supervision and management, performing such tasks as connection and disconnection. This channel operates at a rate of 16 KB/s and is called the delta channel, or D-channel for short. The arrangement of three channels in this manner is referred to as 2B + D and yields an aggregated data rate of 144 KB/s. It is possible to add multiple BRI devices (up to eight) to the same line using the S-bus interface of basic rate ISDN (Telecoms Corner, 2000).

Variations are apparent in the international implementations of the PRI channel structure, amongst others, where the North American implementation has a PRI channel structure of 23 B-channels plus one 64 Kbits/s D-channel, giving an aggregated data rate of 1536 Kbits/s. In most locations, however, it is possible to enter an entirely customized arrangement for whatever bandwidth is required.

In Europe, on the other hand, PRI consists of 30 B-channels plus one 64 Kbits/s D-channel for a total of 1984 Kbits/s. It is also possible to support multiple PRI lines with one 64 Kbits/s D-channel.

H-channels provide a way to aggregate B-channels. They are implemented as:

Note that, in ISDN terminology, ‘K’ means 1000 (10³), rather than 1024 (2¹⁰) as mentioned in Chapter 1; therefore a 64 Kbits/s channel carries data at a rate of 64 000 bits per second. Where confusion arises, the reader is encouraged to clarify.

Acceptance and Administration

Although available for some time, telecommunications companies have been slow to implement ISDN, especially at the local level. One reason for this is the different practical implementations of the original recommendations, which proved to be incompatible between different vendor networks.

The CCITT began the process of standardization of digital telecommunications and ISDN is documented in CCITT Recommendation I.120 (1984) which describes some initial guidelines. CCITT is now known as the International Telecommunications Union or ITU, and is a United Nations organization responsible for co-ordinating the standardization of international telecommunications (International Telecommunications Union – see Notes and further reading).

During the early 1990s, an industry initiative to establish a more specific implementation for ISDN in the United States began. Members of the industry agreed to create National ISDN 1 (NI-1) so that customers would not have to know the specific implementation of the standard within the switch they are connected to in order to install compatible connection equipment. Once the NI-1 agreement was finalized, a more comprehensive standardization initiative, National ISDN 2 (NI-2), was undertaken. Some manufacturers of ISDN communications, such as Motorola (see Notes and further reading) and US Robotics (http://www.usr.com/) have worked to develop configuration standards for their own equipment.

ISDN allows multiple digital channels to be operated simultaneously through the same regular phone wiring used for analogue lines, provided that the lines in question meet a certain criteria. The change between analogue and digital transmission on a particular line comes about when the telephone company’s switches are made to support digital connections. Although the same cable is used, it is a digital signal instead of an analogue signal which is transmitted down the line. This scheme permits a much higher data rate than analogue transmission using traditional modem technology to modulate a signal onto the frequency band used by voice on the telephone cable. In addition, the amount of time it takes to set up an end-to-end conversation between two devices over ISDN is about half that of an analogue link. This has applications for services requiring sustained circuits infrequently, such as interactive applications such as games, video and longer file transmission periods that might be experienced during audio transfer between studios.

The advantage of ISDN when applied to a normal telephone connection is the availability of additional digital services, such as caller ID and multiple lines, with the ability to route calls to different destinations in different ways.

The digital signal also contains information on the caller’s number and the type of call (for instance, data, voice, or fax) as well as other enhancements.

Customer Equipment

The telecommunications provider supplies BRI customers with a two-wire link from the telephone switch known as a U interface, covered in more detail in the section ISDN line types. The connected device is called a network termination 1 (NT1), which converts the two-wire U interface into a four-wire interface called the send–transmit (S/T) interface. Singularly, S and T are electrically equivalent to each other and the S/T interface can support up to seven devices on the bus. The NT1 device may be in the form of a network interface, allowing support for Ethernet connection, for instance.

ISDN devices pass through a network termination 2 (NT2) device, to convert the T interface into an S interface, and most NT1 devices include NT2 in their design. The NT2 communicates with terminal equipment, and handles the data link and network layer protocols. A basic rate installation of 2B + D is known as ISDN2, as two bearer lines of 64 Kbits/s are supplied.

Devices designed to interface with ISDN are known as terminal equipment 1 (TE1) devices, and have an ordinary telephone interface (known as the R interface).

8.3.2  Narrowband ISDN

Narrowband ISDN offers transport data rates in the order of 1.544 MB/s or less. This capacity has been configured in several general service offerings discussed below.

Circuit Switched Voice

This is a digital voice service offered over a four-wire ISDN digital subscriber line (DSL), with customer equipment normally located inside the premises.

Circuit Switched Data

Circuit switched data services provide point to point digital data connection over the public network (see VPN). ISDN uses separate signalling channels for the establishment and maintenance of data connections requiring special processing (see ATM). The customer equipment will normally be purchased as part of the cost of installation and may be retained by the customer. Service agreements cover responsibility thereafter.

Low Speed Packet

A monitoring capability is provided by using the D-channel on DSL. The D-channel provides a data rate in the region of 16 KB/s using the X.25 protocol within the upper layers. The D-channel can be used for low speed packet data, whilst also relaying call-processing information. This can be aggregated with the highspeed packet service to obtain the maximum data rate.

High Speed Packet

High speed packet service describes the configuration of one or both of the B-channels in 64 Kbits/s partitions used for circuit switched voice, circuit switched data, or high speed packet service in a user definable combination.

8.3.3  Broadband ISDN Service

Broadband ISDN service can supply data rates beyond the 1.544 MB/s limit of narrowband, and is usually run on a frame relay, SMDS, or ATM connection-based network. The data rates available for broadband ISDN services range from 25 MB/s up to gigabits per second, and are implemented within backbones.

SONET

Synchronous optical networks (SONET) are based on fibre media for physical layer transmission. As a physical transmission mechanism, standardized SONET technology supports flexible transport architectures. The two most common speeds of broadband ISDN running over SONET type networks are 155 MB/s (OC-1) and at 622 MB/s (OC-3), made possible by the high quality of the digital facilities in place on the network.

Frame Relay

Frame relay supports the transport of data in a connectionless service, meaning that each data packet passing through the network contains address information. Common frame relay services start at 56 Kbits/s and 1.544 Mbits/s, scaling to 25 MB/s and beyond. One unique facet of frame relay service is the ability to support variable size data packets. This has applications for variable bit rate (VBR) data flows, such as might be utilized within interactive applications and simulation.

Switched Multimegabit Digital Service

Switched multimegabit digital service (SMDS) is a digital service providing a high speed digital path for permanent virtual circuits (see VPN). Common service offerings provide data rates from 56 Kbits/s to 34 Mbits/s and up to 155 Mbits/s and beyond, making it scalable from the customer’s perspective. As with SONET, the link is transparent to upper layer protocols, meaning that the functionality of the upper layers can be accommodated. So, for instance, using the TCP/IP protocol suite over SMDS allows virtual subnetworks to be implemented.

8.3.4  Physical Layer

The ISDN physical layer is specified by recommendations published in the ITU I-series and G-series of documents (http://www.itu.ch/publications/itu-t/itutg.htm). Around 800 documents are contained within the G-series and so the full list is not reproduced here! The documents are available at the ITU website, where each descriptive title links to a download of the full text. As an example of the wide scope of the recommendations, one section of the list is entitled ‘General recommendations on the transmission quality for an entire international telephone connection’ and this section contains recommendation G.111 (3/93) loudness ratings (LRs) in an international connection, amongst others.

Local Loop and Backbone

The connection between the customer equipment at the line of demarcation and the telephone switch is called the local loop. Because of the many miles of cables and the variety of physical locations required to service millions of telephone outlets, the local loop is the most difficult and expensive to install and upgrade.

The local loop connection is called the line termination (LT function), whilst the connection between switches within the phone network is called exchange termination (ET) function. The interface between the local loop and the main telecommunications backbone is known as the V interface.

ISDN Line Types

ISDN installation depends upon the characteristics of the local loop. Before the service provider installs an ISDN line, a series of measurements are taken to ensure that the cable pairs are loop qualified and meet the requirements of ISDN. If the current analogue line does not meet the requirements, a new cable is required.

T-interface

All ISDN lines are equipped with an individual line card in the central telecommunications switch, located at the provider’s premises. ISDN users within approximately 1000 m of the local ISDN switch may be provisioned on a T-interface line. A T-interface line provides a direct ISDN connection to the switch and is a four-wire link. A T-interface line delivers the ISDN signal to the terminal since this is equivalent to having the NT1 equipment housed at the provider’s convenience, although the line is actually connected to a card in the switch.

U-interface

The U-interface supports a longer local loop than the T interface. The U-interface line uses a different type of interface card at the switch, and is supplied as a two-wire interface. The length restriction depends upon the encoding technique and wire gauge (on the 2B1Q encoding technique, explained later, the limit is 9000 m of 33 gauge wire or 42 dB loss at 40 kHz). The U-interface also needs to have an NT1 unit located at the user’s premises.

Z-lines

The Z-card line is used where an analogue line may be required in a location served by predominantly ISDN lines. Z-cards are most commonly used for telephones in environmentally difficult areas or for pay telephones.

ISDN Terminals

Power

All ISDN terminals require a power connection and so ISDN is not available in the event of a power failure. The power is used for the terminal encoding and decoding and to provide signalling tones. Unlike an analogue telephone wherein a ringing generator signal is placed on the line to ring the phone, in ISDN an ‘Alerting’ call control message is relayed to the terminal. The ‘Alerting’ message tells the ISDN terminal to play certain tones. The dial tone originates within the ISDN terminal.

Encoding Techniques

As mentioned earlier, the difference between ISDN standards across national boundaries causes incompatibility when interconnecting between implementations. In the physical layer, an example of this can be found within the encoding mechanisms for representing binary states as voltages on the cable. BRI and PRI are different even in the same country, because of the different work that each link type performs, but there are also differences between BRI services between countries.

In America, the BRI service uses the 2B1Q scheme to represent data as voltages on the cable. Europe, on the other hand, uses a technique called 4B3T for BRI services. Both schemes are used in mechanisms achieving higher data rates than are currently available for LANs.

2B1Q

2B1Q is detailed in the 1988 ANSI specification T1.601 which describes a scheme whereby the input voltage level can be one of four distinct levels. As illustrated in Figure 8.2, a level can represent two binary states, since there are 2² possible variations. Note that a voltage level of zero is not valid under this scheme. The figure shows the voltage levels and the pairs of bit states that each represents.

4B3T

4B3T uses return-to-zero (rather than NRZ) encoding to represent 4 bits rather than 2 bits, within one voltage state. Because of this relation to one voltage state equalling 4 bits, the voltage state is called a quaternary.

4B3T is defined in European Telecommunications Standards Institute (ETSI) ETR 080, Annex B and other national standards, such as Germany’s 1TR220. 4B3T can be transmitted reliably at up to 4.2 km over 0.4-mm cable, or up to 8.2 km over 0.6-mm cable.

PRI encoding varies depending upon use, and research is continuing to improve encoding techniques (Telecoms Corner, 2000). In North American PRI, transmission of T1 grade service normally occurs over twisted-pair cable, using one of two types of line coding – AMI or B8ZS. Line coding is defined in ITU recommendation G.703 and a variety of US standards.

AMI Line Coding

Alternate mark inversion (AMI) is used in North American T1 (1.544 MB/s) implementation. With AMI, a positive or negative pulse is generated with each mark or 1. The pulse polarity depends upon the polarity of the last preceding pulse. When a space or 0 is transmitted, there are no pulses generated. Repeaters usually require pulse transitions to recover and regenerate timing on the line, requiring that long strings of 0s be avoided. The synchronization method employed within ADAT uses a similar rule but AMI specifies that no more than 15 consecutive zeros are allowed.

B8ZS Line Coding

Bipolar 8 zero substitution (B8ZS) is a newer line-encoding scheme used within North America. AMI timing recovery problems encountered with consecutive zeros, is solved with B8ZS coding. One of two special bipolar violation codes is substituted for strings of eight zeros. Intentional bipolar violations are caused in bit positions 4 and 7 of the data stream. With B8ZS encoding, no more than seven consecutive zero voltage states will be transmitted on the line. ADAT also uses a deliberate violation pattern in order to facilitate frequency selection.

The violation codes cause two bipolar violations of opposite polarity in the bit stream, which also thus have the effect of preventing excessive DC voltage levels.

8.3.5  Data Link Layer

The ISDN data link layer is specified by the ITU Q-series documents Q.920–Q.923. Apart from the encoding previously described, the remainder of the layer concerns itself with synchronization and error checking.

Frame Format

A US T1 frame consists of 193 bits: 192 bits contain 1 byte of data from each of the 23 B-channels and 1 byte of data from the D-channel (8 bits × 24 bytes = 192 bits). A single bit is prefixed to the data as a stop-bit (or point of reference) for the synchronization.

Frames can be combined into a super-frame or an extended super-frame, which are made up of 12 and 24 frames, respectively.

A European E1 frame consists of 256 bits: 240 of these are used to transport 1 byte of data for each of the 30 B-channels plus the D-channel and 1 byte for framing and synchronization purposes. Sixteen frames are combined to form a multi-frame.

Synchronization

The synchronization pattern within the BRI frame contains the nine quaternaries DC pattern, followed by 12 occurrences of data, aggregated in the form B₁ + B₂ + D. Each aggregation is 18 bits in length, made up of 8 bits from each channel. Finally, the frame is completed with a maintenance field, consisting of redundancy checks, error flags, and sysex commands used for loop-back testing without disrupting user data.

When bundled together in a super-frame, the sync field of the first frame is made significant by inversion of the DC phase, and this represents a string of six consecutive 1s (or 7Eh). 7Eh is known as the flag character and is part of the header information.

8.3.6  Network Layer

The ISDN network layer is also specified by the ITU Q-series documents Q.930–Q.939.

Addressing

Basic addressing occurs in the network layer, and this is contained within the call reference field. The call reference field consists of 2 or 3 bytes. BRI systems use a 7-bit call reference value (with 128 possible combinations) within 2 bytes. PRI systems have a 15-bit call reference value (yielding 32 767 possible combinations).

The call reference value is an arbitrary number, which represent a conversation, or logical link, between the two end devices, allowing devices to manage multiple conversations on the same cable.

Figure 8.3 shows a cell-like network of routers. The value in the call reference field between points A and B will not necessarilybe the same as between points B and C, but serves as an index number that both nodes can refer to when managing connections.

Message Types

There are four general message types: call establishment, call information, call clearing, and miscellaneous. Obvious candidates for inspection are call establishment and call clearing messages. The full list of bit pattern assignments is shown in Figure 8.4. The message type byte identifies the exact message and thereby determines what additional information is required and allowed.

Call Set-Up

In order for a communications channel to be established, the initiating device first sends out a set-up (00101) signal across the D-channel. Under normal circumstances, the switch at the opposite end of the cable, labelled as point B in Figure 8.3, sends a call proceeding signal (00010) back to the initiator, as well as a set-up signal to the intended recipient.

Upon receipt of the set-up message, the recipient device (C) sounds the ring of the telephone and sends an alert signal (00001) back to the switch. This is then forwarded to the initiator (A). When the call is answered, device C sends a connect message (00111) to the switch which is then forwarded to device A. An acknowledgement message (01101) is then returned and the call can proceed.

8.3.7  Audio Implementations

The broadcast industry is able to make use of equipment, allowing the ability to talk off-air with a caller at the same time as other callers are on-air. Such equipment might also include some DSP before the compressing the audio, such as through MPEG Layer III, for better data density. Although some audio artefacts may be lost during such compression, the improvement in transmission efficiency is considered appropriate. Compression technologies are important to the study of audio transfer, although such is the complexity of the subject that further study is recommended!

MPEG Layer III encoding is used to offer full-duplex 15 kHz stereo audio or simplex streamed audio transmission, for which specialist equipment must be purchased.

Applications for this technology might include Internet audio, remote broadcasting, video conferencing, or logging.

8.4  Presto

Presto is a data link interface designed to map audio, video, MIDI, Ethernet, and synchronization data onto the ANSI T1.105–1995 synchronous optical network (SONET). Presto can be implemented as a local interface on a PC board, a point to point interface, a local area network, or a wide area network. Presto can transmit more than a thousand audio channels over fibre optic cable for distances up to 80 km. Presto supports ANSI SONET STS-3c/OC-3 (155 Mbits/s), STS-12/OC-12 (622 Mbits/s), and STS 48/OC-48 (2.49 Gbits/s). It can also be transmitted at OC-3 speeds using category 5 cable among other media types.

There is room for expansion with fibre optic cable lengths of 40 km and longer and speeds of 9.95 Gbits per second and greater.

8.4.1  Standards and Administration

Presto is an initiative undertaken by Kurzweil Music Corporation with revision 0.4 of the specification made available shortly after its publication date of 4 September 1999 (Professional Audio via Synchronous Optical Networ. – see Notes and further reading).

8.4.2  Timing Considerations

Like SONET, Presto is synchronous in nature. Analogue conversion clocks are synchronized to the interface clock rate or vice versa. Various audio sample rates are supported, from 22.05, 24, 32, 44.1, 48, 64, 88.2, 96, 128, 176.4, 192, 256 kHz, 2.8224 and 5.6448 MHz, and sample rates with rational fraction relationships to the 8 kHz SONET frame rate. These rational fractions are used to synthesize conversion clocks from a clock source. The clock is manipulated using a 24-bit number, allowing fine adjustment of sample clock frequency. Sample frequency can also be adjusted without requiring additional buffering, or other causes of delay. Sample word widths of 1, 16, 24, and 32 bits are supported. The maximum capacity of audio-only transport ranges from 128 channels at 48 kHz, 24 bits using STS-3c to 2048 channels at 48 kHz, 24 bits using STS-48.

8.4.3  Data Format

In order to reduce delays and buffer requirements when interfacing with ADAT, TDIF, and other audio interfaces, data are transmitted most significant bit first, and multiplexed as pairs of channels (a/b) or as single channels. Bit reversal within samples is required when interfacing directly to AES3. The most significant data byte within a channel is sent first and is numbered byte 1. The next most significant byte sent next is numbered byte 2, and so on.

Audio is multiplexed by channel, with the most significant byte of the 1st channel transmitted first, the most significant byte of the 2nd channel second, and so on. Once the last byte of the last channel in the last set has been transmitted, the payload is padded.

8.4.4  Error Correction

To decrease the chance of an interface error causing an audible malfunction, data can be transmitted in duplicate. If a CRC error indicates that one copy of the data may be bad, the other copy can be used instead. To accomplish this, at rates of 96 kHz and below, data are transmitted in pairs at twice the normal rate, with channel pair sets carrying a single channel of data each. Once all bytes from all channels have been transmitted once, they are transmitted again. There is at least one 16-bit CRC for each copy of data transmitted.

8.5  Asymmetrical Techniques

Asymmetric techniques allow higher data rate in one direction than in the other. From a subscriber’s point of view, this means that lots of data can be received, but fewer data can be sent out in the same time period. Asymmetrical techniques are useful in instances such as video-on-demand, where the data rate coming into a subscriber’s premises is required to be much higher than the outgoing data rate. An imagined video-on-demand service, for instance, might require that a subscriber make a request to view a video at a pre-selected time. The time that the video is transmitted to the subscriber’s premises might be one of several predetermined times made available by the video provider, or transmitted at the subscriber’s request, beginning seconds after the request is made, as described in Chapter 1. In either case, the data sent out from the subscriber whilst making the booking are only a tiny fraction of the data received whilst viewing the video.

Another service, to which asymmetrical techniques are being applied, is fast Internet access. This is possible because of the behaviour of the vast majority of users of the Internet, where a relatively small address is sent as a request for a larger amount of information.

8.5.1  ADSL

Asymmetric digital subscriber line (ADSL) is a subscription-based service used to deliver higher rates than BRI ISDN over the existing POTS infrastructure.

ADSL is one member of a family of transport systems called digital subscriber line (xDSL) and practical implementations are capable of offering data rates between 1 and 10 MB/s over copper lines. Outgoing data rates from the subscriber’s premises are in the region of 64–640 Kbits/s.

ADSL was motivated by a requirement to make available the use of a regular telephone connection in the event of a power failure, offering an improvement in service over BRI ISDN. In general, the fastest DSL interfaces can only be supported within 2 km of the switch, although the implementations offering slower data rates can go further.

General Description

Subscription to an ADSL circuit requires an ADSL modem on each end of a twisted pair telephone line. Three communication channels are created on the cable, and these are split into a high speed downstream channel, a medium speed duplex channel that depends on the implementation of the ADSL architecture, and a POTS or ISDN channel. The POTS/ISDN channel is removed from the digital modem by filters allowing uninterrupted POTS/ISDN, in the case that the ADSL portion fails. The high speed channel ranges from 1.5 to 8 Mbits/s, whilst the duplex rate ranges from 16 to 640 Kbits/s. Each channel can be subdivided to form multiple, lower rate channels.

Dividing the available bandwidth of a telephone line in one of two ways creates the multiple channels. Frequency division multiplexing (FDM) is used to assign one band for upstream data and another band for downstream data. The downstream path is then further divided into one or more high speed channels and one or more low speed channels. The upstream path is also multiplexed into low speed channels. With either technique, ADSL splits off a 4 kHz region for POTS.

The implementation of FDM is of interest because it is this that allows ADSL to be delivered through the various quality cables that make up the POTS infrastructure. In more detail, frequencies are divided into 4 kHz blocks, each one becoming a data channel in its own right, with POTS making up a specific frequency range. Where poor quality cable impedes the transmission of a certain channel, ADSL has the ability to cease transmission on that channel, and adjust and manage the data rate accordingly. Some configurations of the POTS infrastructure, such as the presence of line amplifiers with certain electrical properties on the link, can completely obstruct all channels, meaning that ADSL cannot be usefully used to transmit data. Customer experience of ADSL may vary, depending on factors including the distance from the provider’s premises and the quality of the available POTS line.

Another advance which went into ADSL was in the error correction mechanism. As a real-time signal, digital video cannot use link- or network-layer error control, for reasons previously explored. However, ADSL incorporates a forward error correction mechanism which reduces errors caused by impulse noise, thereby removing a significant cause of errors. This technique introduces a delay in the order of 2 ms.

The data format within a channel is again frame based, with an error correction code attached to each frame or block. The receiving modem corrects errors that occur during transmission up to the limits implied by the code and the block length. Optionally, the unit may also create super-blocks by interleaving data within sub-blocks, thereby allowing the receiver to correct any combination of errors within a specific span of bits. Typically, an error rate of 1 in 10 000 000 bits or higher is achieved, allowing for effective transmission of both data and video signals. Video compression technology allows transmission requiring a data rate of around 1 MB/s for VCR quality and 2–3 MB/s for broadcast quality pictures (Direct TV satellite systems use approximately 3–4 MB/s. Source: ADSL Forum, see Satellite). Practical implementations have achieved data rates beyond the originally anticipated 6 Mbits/s and led to the transmission of at least one such video channel or a number of audio channels (ADSL Forum, 1998).

Because of the wide variety of environments characterizing the local loop, problems occur in the implementation of digital technology. Long telephone lines may attenuate signals at 1 MHz (the outer edge of the frequency range band used by ADSL) by up to 90 dB. Developments in frequency splitter technology have allowed the realization of a large dynamic range, with low noise figures.

ADSL can be purchased with various speed ranges and capabilities. The minimum configuration provides 1.5 or 2.0 Mbits/s downstream and a 16 Kbits/s duplex channel. Improvements continue to increase the upper data rates available to subscribers.

Standards and Associations

Although dispute regarding line coding techniques (Discrete Multiton … – see Notes and further reading) prevented early adoption, ANSI administrates the standard and technical recommendations within the North American continent. Within Europe, administration is performed by the European Technical Standards Institute (ETSI – see Notes and further reading).

The ANSI working group T1E1.4 approved an ADSL standard for rates up to 6.1 MB/s (ANSI Standard T1.413) (Alliance for Telecommunications Industry Solutions – see Notes and further reading). ETSI contributed an Annex to T1.413 to reflect European requirements. T1.413 currently embodies a single terminal interface at the subscriber’s premises. Subsequent issues include a multiplexed interface with additional recommendations for configuration, network management, and so on. The ATM Forum and the Digital Audio Visual Council (DAVIC – see Notes and further reading) recognize ADSL as a physical layer transmission protocol for unshielded twisted pair media.

DAVIC was created in August 1994 with an expected duration of 5 years. Its task was to create complete sets of specifications using emerging digital audio–visual technologies. The term of 5 years was reached in August 1999 and the set of specifications (DAVIC1.3.1) had become an International Standard and International Report (IS 16500 and IR 16501) after the resolution of comments performed in June 1999.

The ADSL Forum develops technical guidelines for architectures, interfaces, and protocols for telecommunications networks incorporating ADSL transceivers. The ADSL Forum was formed in December 1994 to represent commercial organizations involved with ADSL and to lend the weight of practical implementations to such areas as system architectures, protocols, and interfaces. In November 1999, the Forum voted to change the name, by dropping the A. It felt that the adoption of the new name would demonstrate more clearly that the Forum is an all-embracing worldwide body, covering all the varieties of DSL, and the name was subsequently changed in January 2000 (DSL Forum – see Notes and further reading).

Physical Layer

Cable

As mentioned, the performance or availability of ADSL to particular premises relies on the proximity to the telecommunications facility, or in other words, the length of the local loop. The table in Figure 8.5 indicates the data rate/distance ratio and the types of cable expected. The cable type is twisted pair.

ANSI T1.413 specifies the frequency range of 26 kHz to 1.1 MHz, with frequencies below 4 kHz reserved for POTS. This means that the use of a normal telephone over an ADSL connection will not affect the data rate of the connection.

Load coils used on voice lines to improve the quality of the voice service do not allow signals above the voice band 0–4 kHz to pass through them, meaning that ADSL is not available on such lines.

Encoding

Implementations of DMT within ADSL divide the downstream channels into 256 4 kHz wide tones and the upstream channels into 32 sub-channels. Data are allocated onto a sub-channel and one sub-channel transmits a stream of data serially. In this way, a parallel transmission is achieved. Transmission of data is turned off, for instance if a particular sub-channel is prone to RF interference over a particular connection.

Combining ISDN and ADSL Signals

The methods for combining ISDN and ADSL (Transmission and multiplexin. – see Notes and further reading) can be placed into the general categories of in-band and out-of-band, and these are illustrated in Figure 8.6.

In-band combinations treat the ISDN signal as another stream of data to be carried by ADSL. This approach has the advantage that the ADSL modem is free to use the standard frequencies for transmission and start-up. Using this method allows compliance with the T1.413 standard, but means that the ISDN signal, including the voice component, is processed by the ADSL modem. Since the modem requires local power, a local power failure will also cause the lifeline telephone service to fail. Furthermore, the processing delay of around 2 ms, introduced by the forward error correction to compensate for impulse noise, is more than the maximum delay of 1.25 ms tolerated by ISDN.

Out-of-band combinations leave the ISDN signal intact, to be transmitted separately at its usual frequencies. The ADSL signal turns off channels that encroach upon ISDN bandwidth, thereby restricting ADSL to higher frequencies. This technique requires a departure from the T1.413 standard, although the benefits of the out-of-band method are substantial and include the lifeline telephone connection, thereby justifying the required deviation from the standard within practical implementations. A passive filter is required to correctly separate the ISDN and ADSL signals from one another (Orckit 38 – see Notes and further reading).

Applications

ADSL has been applied to video-on-demand, and this is often enhanced with additional consumer-oriented applications taking advantage of the available data rate in each direction. Another reason for the interest amongst consumers has been for fast Internet access whilst protecting the investment in POTS. Since ADSL can use large proportions of the copper infrastructure, and no additional cables need to be laid in the majority of areas, the service is a cheap way to increase the inbound data rate.

In practice, the benefit depends upon the distance from the switch, and may be completely impeded by line-driving equipment. In short, customers’ experiences have been varied in the early stages of deployment.

A practical application is the DSL-Lite initiative, which envisages a USB interface within the ADSL modem. Administration for DSL-Lite is performed by the UAWG and ITU.

8.6  Satellite Communications

The use of satellite communications as a consumer technology really started at the end of the 1990s with the retasking of redundant military satellite hardware on a commercial scale, following the end of the cold war.

From the consumer perspective, transmitting to a satellite is expensive, because of the high powered equipment required for the job, and the financial arrangements with the satellite owners. However, it is relatively cheap to receive information transmitted from the satellite, and a simple example of this can be seen in the prevalence of satellite television such as Direct TV in North America and parts of Europe. Satellite television is very much in the broadcast industry sector and subscription involves a small dish installed at the consumer’s premises. In this way, large amounts of data can be received once the dish has been correctly aligned with the satellite. Satellite communications facilities offer various services, with Direct TV able to supply a selection of 200 video channels.

Apart from television, the other use to which this kind of communication has been put is fast Internet access. In order to overcome the practical problems of two-way communication through a satellite, the outgoing data are handled by transmission through a link such as ISDN BRI, or telephone modem, to the operator of a satellite transmission premises. The uplink transmits the data from hundreds or thousands of sources through a large dish or base station. The satellite then bounces the data back down to the ground for receipt by the consumer’s dish, thereby completing the loop.

Service offerings are scalable and bandwidth can be supplied on demand or as a commodity. For instance, if a television broadcast is required, bandwidth can be purchased on the link in order to do this, with the bandwidth being reallocated once it is no longer required. Such a system might be adopted for the broadcast of business television.

Satellite communication can be made global by bouncing the signal from satellite to satellite, thereby overcoming line-of-sight problems encountered due to the curvature of the earth (see Figure 8.7).

Shown in Figure 8.8, satellite communication is associated with a footprint, since one geostationary satellite is able to transmit permanently to a certain area, depending upon its location. Orbit types can be classified in a number of ways, such as high and low orbit, and geostationary or geomotionary. High orbit satellites have a longer latency incurred simply because of the distances that the signal must travel to and from the satellite, but high orbit satellites also have a larger footprint. Geostationary satellites move at the same speed as the rotation of the earth and therefore maintain a constant presence in a particular position in relation to the ground. These types of satellite are more generally suitable for the provision of telecommunication because of their static nature.

8.7  IBM Network

The IBM network is chosen as an example of a privately owned global network by virtue of one of the first global data networks to be used commercially. The International Business Machines Corporation (IBM) built the network principally for the transmission of the proprietary systems network architecture (SNA) protocol. More recently, the network has been made capable of transmitting IP, with the result that it has become a significant part of the Internet. This is true of only a handful of commercially owned networks.

Connection to the IBM network must be made by subscribing to the appropriate Internet provider, or by purchasing a telecommunications link to the nearest network node. Network nodes are strategically placed internationally, covering many major cities.

On 8 December 1998 AT&T and IBM announced a series of agreements under which AT&T acquired IBM’s global network business for $5 billion in cash. The contract for the US network was signed on 30 April 1999, and subsequently in other countries.

Various service offerings include remote access services, IP remote access, DSL and Internet VPN gateway and global managed Internet service (AT&T – see Notes and further reading), such as was used for the IEEE 1394 over long-distance tests, mentioned in Chapter 5.

IBM offers consumer level Internet access over the same network. From the consumers’ point of view, globally managed network provision uses the same cables as the Internet access, although the bandwidth is guaranteed for the entire period of the contract, and so it is possible to purchase very high data rates permanently attached across the globe.

Notes and Further Reading

ADSL Forum (1998) Frequently Asked Questions. DSL Forum Office, 39355 California Street, Ste. 307, Fremont, CA 94538, USA. http://www.adsl.com.tech_faqs.htm.

Alliance for Telecommunications Industry Solutions. 1200 G Street, NW, Suite 500, Washington DC 20005, USA. http://www.t1.org.

Appleby, S. (1994) BT Technology Journal, April, pp. 19–29. British Telecom Res. Labs, Ipswich, UK.

AT&T. 295 North Maple Avenue, Basking Ridge, NJ 07920, USA. http://www.att.com/globalnetwork.

Digital Audio Visual Council (DAVIC), Geneva, Switzerland. Only available at http://www.davic.org.

Discrete Multitone (DMT) vs. Carrierless Amplitude/Phase (CAP) Line Codes (2000) Aware Technologies, 40 Middlesex Turnpike, Bedford, MA 01730, USA. http://www.aware.com/technology/whitepapers/dmt.html.

DSL Forum. 1212 Suffolk Street, Naperville, IL 60563, USA. http://www.adsl.com.

European Telecommunications Standards Institute (ETSI), 650 Route des Lucioles, F-06921 Sophia Antipolis Cedex, France. http://www.etsi.org.

International Business Machines Corporation. New Orchard Road, Armonk, NY 10504, USA. http://www.ibm.com.

International Telecommunications Union (ITU), Place des Nations CH-1211 Geneva 20, Switzerland. http://www.itu.int/.

Internet2 International. University Corporation for Advanced Internet Development, 1112 16th Street, NW Washington DC 20036, USA. http://www.internet2.org.

Motorola Incorporated. European Headquarters, Church Road, Lowfield Heath, Crawley, West Sussex, RH11 0PQ, UK. http://www.mot.com/.

Orckit 38 Nahalat Yitzhak Street, Tel Aviv 67448.

Professional Audio via Synchronous Optical Network (Presto) (1999) Rev: 0.4. Kurzweil Music.

Telecoms Corner (2000) Technical Reference site. http://tele-com.tbi.net/.

Transmission and multiplexing (TM); access transmission systems on metallic access cables; asymmetric digital subscriber line (ADSL) – coexistence of ADSL and ISDN–BA on the same pair. ETSI TS 101 388.