Basic Concepts of Communications: An Introduction
24 September 2003
Richard A. Costello, Ray Horak
Document Type: Technology Overview
Note Number: DPRO-89990
To develop a solid understanding of communications technology, one must be firmly grounded in a wide range of basic concepts in both the voice and data domains.
Table of Contents
List of Tables
List of Figures
Technology Basics [return to Table of Contents]
Telecommunications is defined as the transfer of information over a distance. The information can be audio (for example, voice), image, video, computer data or any combination. The voice or video information can be transferred in its native analog format over an analog network, or it can be transformed into a digital format for transfer over a digital network. Similarly, computer data can be transferred in its native digital format over a digital network, or it can be transformed into an analog format for transfer over an analog network.
In the voice and video domains, the telephones accept acoustic analog audio inputs, and the cameras accept reflected analog light inputs on the transmit side of the data transfer, and telephones and TV monitors recreate them on the receive side. In the domain of data communications, the data may take the forms of text, graphics and images in a wide variety of formats including documents, spreadsheets, e-mail, customer records, performance records, traffic tallies, still images and video. The originating and destination computing devices can be dumb terminals, PCs, high-performance workstations, minicomputers or mainframes.
The network connecting the terminal devices may be either analog or digital in nature or a combination of the two. It may be a dedicated network, connecting the terminal devices directly, or it may be a switched network. The network may operate over a local area, a metropolitan area or a wide area. The network may be may relatively simple in nature or quite complex, perhaps comprising a great number of network elements, the most prevalent of which are transmission and switching systems. There may be a great many switching devices interconnected by a great many circuits, thereby offering a considerable number of alternative physical routes or paths for the data to take as it moves between the communicating terminal devices.
Prerequisites for Communications
Specific conditions must be met before communications can occur. Regardless of the nature of the information exchange, certain standards for conduct must be understood, agreed on and followed carefully. For example, two people attempting to exchange ideas must agree on a common language for communications to take place. Otherwise, no communication occurs, even though words are spoken by one and heard by another. The communication is more complete and effective if the two people can see each other, for facial expressions and body language can be quite meaningful.
Certain rules and conventions must be followed in order that the communication is as effective as possible. For example, if the communication takes the form of an interactive conversation rather than a lecture, the two people must take turns talking, and the listener must be attentive. The voices must be modulated so that the volume and frequency levels are acceptableit is difficult to hear a whispered word, yelling and screaming can be very unpleasant for the listener, and a monotone may put the listener to sleep. The pace of the speaker must take into consideration the ability of the listener to absorb and understand the data transfer. If there is excessive background noise, the volume levels may have to be increased, the pace may have to be slowed and words may have to be repeated in order to compensate for communications errors. These rules and conventions for communications can be characterized as protocols.
Data communications networking involves a great many protocol considerations. Devices must also speak the same language and follow the same rules, and mechanisms must be in place to ensure that data travels from one device to another without errors. Unfortunately, commercially available data communications devices speak a variety of tongues and follow a number of different rules, thereby causing real confusion among data communications users.
However, various organizations, including the International Telecommunication Union-Telecommunication Standardization Sector (ITU-T), the International Organization for Standardization (ISO), the Electronic Industries Alliance (EIA) and the Institute of Electrical and Electronics Engineers (IEEE) have developed standards that are widely recognized and used throughout the industry in the development and deployment of networks, and in their interconnection.
Open Systems Interconnection
To communicate at all, and certainly to communicate effectively, devices must be compatible on various levels. The ISO's Open Systems Interconnection (OSI) reference model consists of a set of international networking standards known collectively as X.200. The model defines a set of common rules and conventions that computers of disparate origin can use to exchange information (that is, communicate). The model is organized into seven distinct layers in order to segment software responsibilities. Although the OSI model does not describe a specification for any particular network or network element, it does serve as a reference point for the establishment of a standard data communications system.
Each layer of the OSI model defines a particular function involved in the transfer of data from one machine to another. The model is structured in an upwardly compatible manner. If there is compatibility on one level, it is assumed that compatibility exists on the levels beneath. These levels are briefly defined in figure "The OSI Reference Model."
[return to List of Figures]
The OSI Reference Model
The ISO OSI reference model defines a seven-level hierarchy for data communications. Depending on the protocol, certain levels will not be implemented.
In very basic data communications configurations, the prerequisites are relatively few and involve compatibility on only the first two layers of the model. In support of end-user applications across complex networks, however, hundreds of conditions must be met before communications can occur, and compatibility must be established on all seven levels.
A physical interface (layer 1) defines the physical connection between two network elements, specifically, between a device and a transmission medium, such as twisted pair, coaxial cable or optical fiber. It is comparable in some respects to an electrical plug-and-socket connection with "male" and "female" components. Data and control lines from a device terminate in a connector with pins that handle assigned signal functions, such as carriage return, line feed and request to send (information). For example, EIA-232, more commonly known as RS-232, is the industry-standard interface for connecting data terminal equipment (DTE) and data communications equipment (DCE). Communications hardware and software drivers also are found at this level.
Each pin in a 25-pin connector represents a standard specification. Pin assignments are explicit and unalterable, except for those that are unassigned. Unassigned pins can be used to handle special functions, such as "busy out," on a modem, a condition that causes a modem to go "off the hook." V.24, another common data communications industry standard, is functionally compatible with RS-232 and RS-449. It specifies standards for expanded transmission speeds, longer cable lengths and additional functions.
Universal Serial Bus (USB) is a 12-Mbps serial bus developed by several leading PC and communications vendors to facilitate a "plug and play" capability for devices (for example, printers, modems, mice, telephones and joysticks) that connect to the communication ports on new PCs. The idea behind "plug and play" is to have the PC automatically detect and configure any devices that are connected to its ports. USB also includes a "hot attach" capability, which enables devices to be connected, detected and configured (and disconnected) even when the power is on.
Information is created, stored, output from and input to computers in binary format, that is, 1s and 0s, each of which is known as a bit (binary digit). A coding scheme, which resides in layer 6 of the OSI reference model, defines a set of characters, including alphanumeric characters, punctuation marks and control characters. Each character in the set comprises a unique bit pattern of a standard length known as a "byte format." A byte commonly comprises an octet of eight bits, although some coding schemes specify byte lengths of four, seven or even 16 bits.
For example, the widely used American Standard Code for Information Interchange (ASCII) is a seven-bit code. Seven bits of 1s and 0s represent each character in the code set. The ASCII code consists of 128 (27) characters, including 95 graphics characters (that is, letters, numbers and punctuation marks) and 33 control characters (for example, tab, escape, backspace, line feed and carriage return). As the common unit for storage and processing is an eight-bit byte, the eighth bit is used in various ways. In a pure ASCII environment, it is used as a parity bit for error controlvery poor error control. In DOS-based PCs, the extended ASCII code set of eight bits yields 28, or 256, unique combinations of bits, each of which is used for a character. Specifically, the additional 128 characters are used for foreign language symbols (for example, Ö, ä and è) and graphics symbols (for example, a, b and m). In Macintosh computers, the 128 additional values are user-definable.
Other commonly used codes are Baudot and Extended Binary-Coded Decimal Interchange Code (EBCDIC). Baudot, named after Emil Baudot and first established in 1874, is a five-bit code used on vintage teleprinter terminals, such as those made by Telex. It does not, however, have an error-checking capability or means of checking that the information is received. EBCDIC, the code used on synchronous IBM equipment, consists of eight-bit coding, which yields 28, or 256, characters. An end-user device that operates with one type of code (for example, ASCII) cannot accept data from a device using a different code (for example, EBCDIC) unless a conversion is performed to make one code compatible with the other. Code converters handle this relatively simple protocol conversion function in software.
Unix- and DOS-based operating systems, except for Windows 2000 and its predecessor Windows NT, use ASCII for text files. These systems use the newer Unicode, a 16-bit coding scheme that yields 65,536 (216) unique character definitions. The value of Unicode is in its ability to support complex alphabets, such as Japanese and Chinese. With over 40,000 defined characters, Unicode can support multiple alphabets simultaneously.
Primarily for historical reasons, however, this largely is an eight-bit world. So Unicode Transformation Format (UTF-8) was developed by the ISO to allow ASCII and Unicode to coexist on the same machine. This is relatively simple, as ASCII matches the first 128 characters of Unicode. Aside from UTF-8 and related UTF standards, various conversion programs allow different operating systems to change a file from one code to another.
Communications protocols cover a wide spectrum and range from single character-by-character transmissions with no error checking to complex rules for moving large amounts of data among many devices. In general, communications protocols comprise three major areas:
Prior to transmitting the data, protocol-specific software in the originating device chunks the data into datasets, or payloads, of specified size and adds overhead bits for various control purposes. The receiving device checks the overhead bits as it receives the payload bits, perhaps checking for errors created in transit. If no errors are detected, the receiving device may send an acknowledgement, indicated that the dataset was received correctly. If, on the other hand, it detects errors, it generally requests a retransmission through a negative acknowledgement.
High-Level Data Link Control (HDLC) is one of the most commonly used layer 2 framing protocols. HDLC encapsulates layer 3 (Network Layer) packet data, adding data link control information. Variations of HDLC are used in X.25 public networks and in frame relay public and private networks.
[return to List of Figures]
The beginning eight-bit flag alerts the receiving device that a frame of data is about to be received. The end flag signifies that the transmission is complete. The address field names the station to which the information is being sent (or in some cases the sending station). The control field identifies the type of frame and provides sequence numbers of frames transmitted and received. The frame check sequence provides an error control mechanism.
Beginning and end flags are actually the same 8-bit sequence (1000 0001). In a typical sequence of frames, there is only one flag between frames, which serves to end one frame and begin the next. If one or more bytes of data anywhere in the frame look like a flag, the receiver will get out of sequence and will be unable to rebuild the payload. To avoid this, the sending end looks at every byte being coded and converts it to a defined "escape" sequence if it detects a flag character in the payload. At the receiver, flag characters perform their task and are stripped away; the receiver then restores any detected escape sequences to the original coding.
Communications protocols are either bit-oriented or byte-oriented:
As there is a great number of protocols, there are significant issues of protocol incompatibility at all seven layers of the reference model. Where these issues surface, protocol conversion must be accomplished in software, which often resides in devices known as "gateways."
There are two types of data transferparallel and serial. Parallel data transfer employs a communications interface with a sequence of dedicated wires, each serving one purpose. A typical parallel communications cable has eight data wires, each of which is dedicated to transmitting a given bit in an eight-bit character byte. Thereby, the entire byte can be transmitted in one strobe, in parallel.
This form of transmission is, however, limited by distance. If one wire is slightly shorter or longer than the others, the data from the eight separate channels will not arrive at the same instant in time and cannot be matched exactly. This phenomenon is referred to as "skew." In addition to the data wires, there typically are a number of wires for various signaling and control purposes. The large number of wires makes parallel transmission impractical over distances greater than a few feet.
Most data communications devices transmit in a serial fashion. Used for long-haul communications, serial data transfer operates bit-by-bit over a single wire rather than in parallel over multiple wires. Therefore, the data must be converted from parallel to serial bitstream prior to transmission. This protocol conversion process is accomplished through a shift register.
[return to List of Figures]
Types of Data Transmission
In a parallel transmission, each bit in a character is transmitted simultaneously on a separate circuit. In a serial transmission, bits are transmitted in sequence over one circuit.
Data transmission can be asynchronous or synchronous in nature. Asynchronous transmission, often called "start-stop transmission," transmits one character at a time. Each character has its own "timing" device (that is, a start bit and a stop bit), which alerts the receiver to the beginning and end of each character. The time between the transmission of each character is referred to as idle time. Asynchronous transmission is inefficient, due to the overhead required for start and stop bits and the idle time between transmissions. With asynchronous communications, transmission speeds as high as 33.6 Kbps can be supported in most equipment today.
[return to List of Figures]
Start and Stop Bits
During transmission, a data character consisting of seven or eight bits is preceded by a start bit and ended with a stop bit that lets the receiving device know where a character begins and ends. A parity bit sometimes is used for error control.
Synchronous transmission sends a set (that is, block, frame, packet or cell) of characters over a communications link in a continuous bitstream. Data transfer is controlled by a timing device called a "clock." Initiated at the sending device (for example, terminal, modem, multiplexer, switch or router), the clock runs at a frequency equal to the transmission rate. Each dataset is preceded by sync bits or a unique character pattern to allow synchronization, and special idle characters are transmitted if no data is being sent.
To accommodate the transmission of a large number of blocks, terminals involved in synchronous data transfer must have buffers for storing the character blocks, as they wait to be transmitted or processed at the receiving end. Synchronous transmission is generally used for higher-speed data transfers. In many cases, a device can operate both asynchronously and synchronously.
Three modes of data transmission are defined: simplex, half-duplex and full-duplex:
[return to List of Figures]
Simplex transmission is unidirectional. Half-duplex transmission is bidirectional, but in only one direction at a time. Full-duplex transmission is simultaneously bidirectional.
Transmission Signal Attributes
All telecommunications systems and networks make use of electromagnetic energy to support the transmission of data. Whether in the form of electricity, radio or light, all electromagnetic energy travels in waves. The signal waveform has three fundamental attributesfrequency, amplitude and phase. Frequency is the number of cycles in a given time, usually a second, and is measured in hertz (Hz). Amplitude refers to the strength of the signal as illustrated by the height of the peaks of the waveform and, in the case of sound, determines the volume or "loudness."
The figure "Transmission Signal Attributes" shows two sinusoidal waveforms in which B lags behind A. They are said to be out of phase with one another. Phase refers to the position of the waveform as it rises and falls and is of particular importance in modulation theory.
[return to List of Figures]
Transmission Signal Attributes
An electromagnetic signal has three fundamental attributesfrequency, amplitude and phase.
Analog vs. Digital
All communications can be characterized as either analog or digital in nature. Analog communications is characterized by the presence of a continuous electromagnetic waveform that varies in terms of frequency and amplitude. Digital communications is characterized by the representation of information in binary form (1s and 0s) and is accomplished by the manipulation of the waveform in a series of pulses or blips of discrete value at specific points in time.
There are a number of ways that the electromagnetic signal can be manipulated, or modulated, to support digital transmission. The simplest approach is On-Off Keying (OOK), whereby the signal is switched on to represent a 1 bit and switched off to represent a 0 bit. The signal is switched at precise points in time and remains on or off for precise periods of time, known as "bit times," so that the receiving device knows what to expect and so that multiple adjacent 1s and 0s can be represented easily. OOK originated in telegraphy and is the technique used in most fiber optic systems. There are a great many much more sophisticated digital signal modulation techniques that involve various combinations of amplitude, frequency and phase modulation.
Bandwidth refers to the information-carrying capacity of a communications circuit or channel. At the analog level, bandwidth is defined as the range of frequencies that the channel is capable of transmitting without interference or signal loss and is measured in hertz (Hz). The greater the range of frequencies a medium can handle, the greater its information-carrying capacity. A broadcast-quality entertainment TV signal, for example, requires bandwidth of 6MHz, whereas a voice signal requires a relatively modest 4kHz.
In digital terms, bandwidth is generally specified in bits per second (bps). A 2-Mbps channel can support a transmission rate of two million bits per second. This rate can be accomplished through simple OOK, as we discussed above, or through sophisticated modulation schemes that involve manipulation of the analog carrier, with "carrier" referring to the frequency or frequency range that carries the communications signal.
Bits and Bauds
Baud is an old term used in the days when teletypewriters were considered cutting-edge terminal technology. Baud rate refers to the number of signal events (that is, signals or signal changes) occurring per second over an analog circuit. In contemporary terminology, baud rate refers to signaling rate of a modem and is roughly equivalent to Hz.
Bits per second (bps) refers to the number of binary bits transmitted per second. The impression of bits on an analog signal is the essence of a digital signal modulation scheme. In a modem communication, the baud rate refers to the signaling rate, while the bit rate refers to the rate of information transfer.
Signaling Rate, Transmission Rate and Throughput
In the digital domain, there are even more subtle distinctions in terminology, which are no less important. A high-level discussion of T1 will serve to illustrate the differences between signaling rate, transmission rate and throughput.
The signaling rate refers, once again, to the number of signal events occurring per second over the circuit. In an electrically based T1 circuit running over unshielded twisted pair (UTP), the signaling rate is 1.544 Mbps. Although the specifics of the modulation technique can vary, the simplest technique involves an analog carrier modulated at a nominal rate of 1.544 Mbaud. Therefore, it can be said that a baud equals a bit in the case of this simple unibit modulation scheme and therefore that the baud rate equals the bit rate.
The actual transmission rate, which refers to the rate of information transfer, is 1.536 Mbps. The difference is 8 Kbps, or 0.008 Mbps, which is attributable to the fact that the signaling rate includes 8,000 framing bits per second. These framing bits are not information bits (that is, payload bits). Rather, they are control bits used variously for purposes of synchronization, circuit monitoring and error control, with the specific uses being sensitive to the T1 framing convention involved. This transmission rate is under the best of circumstances, as older circuit termination equipment requires additional signaling and control bandwidth, which it gains by seizing some data bits and inserting signaling and control bits in their place. This process of bit robbing further reduces that actual data transmission rate, or payload rate.
Throughput refers to the actual amount of good data put across the circuit per second. If there are errors in transmission caused by inherent circuit noise or some external source of electromagnetic interference (EMI), the integrity of some data may be affected, in which case the affected data must be retransmitted, which fact affects throughput negatively.
The quality of the transmission can be thought of as the conveying of information without the loss of information or the addition of any unwanted information (noise). The quality of a transmission system is usually measured in terms of the signal-to-noise ratio, which is usually expressed logarithmically in terms of decibels (dB) and bandwidth. The equivalent measure of the quality of a digital system is the bit error rate (BER).
Multiplexing Techniques [return to Table of Contents]
Multiplexing allows multiple channels to be derived from a single circuit, thereby allowing multiple lower-bandwidth transmissions to share a single higher bandwidth circuit. Although the high-bandwidth circuit generally is more expensive than a low bandwidth circuit, multiplexing allows a single multichannel circuit to be shared among many transmissions. The advantages, therefore, are increased efficiency and lower associated circuit costs. There are two basic approaches to multiplexing:
Frequency Division Multiplexing (FDM)
FDM was the earliest and most fundamental method of multiplexing. It divides the frequency spectrum of an analog circuit into multiple independent, lower-speed subchannels, each of which operates on a separate and distinct frequency band within the available spectrum. The bandwidth of each channel is sensitive to the width of the frequency band. Therefore, the slower the transmission rates, the more subchannels can be derived from the circuit. FDM supports multiple simultaneous transmissions, one per channel, to coexist on the circuit, which is a distinct advantage in many applications.
FDM was the basis for early voice networks but has been replaced in large measure by time-division techniques. FDM continues to find important applications, though, such as in the deployment of Digital Subscriber Line (DSL) technology and in optical fiber media, where the technique is known as "wavelength division multiplexing" (WDM).
[return to List of Figures]
Multiplexing increases the efficiency of communications links, allowing a business to lease a single shared high-speed circuit for less than it costs to lease multiple low-speed circuits.
Time Division Multiplexing (TDM)
Rather than divide a communications link into frequency-separated channels as FDM does, TDM divides time into slices called "time slots." With TDM, each inputting terminal takes its turn at transmitting and receiving data in a continuous fashion. Depending on the multiplexer type, the device accepts only one bit, byte or packet of data from each input line, puts it into a specifically allocated time slot on the high-speed circuit and then moves on to the next terminal in the sequence. The process of accepting data from many terminals in succession is called "interleaving."
At each stage of TDM, the ensemble of groups of bits from respective channels, plus a flag, is called a "frame." A flag is a special pre-defined pattern of bits that indicates the beginning of the frame and enables the receiver to work out which bits belong to which channel. Input signals are sampled one after the other at high speed; only one sample of a specific signal occupies the channel at any one time.
Pulse Code Modulation
Pulse code modulation (PCM) is a standard, fundamental process for converting analog data into the digital format required for transmission over a digital circuit. Standardized by the ITU-T as G.711, PCM involves three steps:
[return to List of Figures]
Pulse Code Modulation (PCM)
The analog waveform is sampled every T seconds. The samples are then converted to integer numbers using eight bits per sample. Finally, the integer numbers representing the samples of a given channel are sent in sequence as a binary bitstream to the TDM process shown in the lower part of the previous figure.
Sampling measures the amplitude of the signal at the given sampling rate in order to extract all the information. According to the Nyquist Theorem and Shannon's Law, the two theorems that define the specifics of PCM, the sampling rate needs to be at least twice the highest frequency of the signal in order to encode the analog signal, send it across a digital circuit and reconstruct a high-fidelity facsimile.
If we take the bandwidth of speech to be 4kHz, then the signal must be sampled at least 8,000 times per second (2×4,000Hz) to preserve the original information for transmission in digital mode. Further, the samples must be taken at a precise pace of 1/8000 second, or every 125 microseconds. Each analog sample is then encoded into an 8-bit binary approximate value through a process known as quantization, based on a table of 255 standard values of amplitude. As this process takes place 8,000 times a second, the basic bandwidth requirement for high-fidelity voice is 64 Kbps.
Digital Transmission Systems (T1 and E1)
T-carrier was designed specifically to support digitized voice communications based on PCM. To achieve backward compatibility with established analog carrier systems, T1 specifies 24 channels. In traditional PCM applications, each channel is 64 Kbps wide, consisting of 8 bits sent 8,000 times per second. Each 8-bit sample is interleaved with 23 others to form a frame of 192 bits. A framing bit is added to indicate the beginning of the frame and, variously, to support synchronization, circuit monitoring and error control functions. The result is a frame of 193 bits. The signaling rate, therefore, is 1.544 Mbps (1,544,000 bits per second). This consists of the aggregate rate due to speech (24 speech channels × 64 Kbps per channel = 1,536,000 bps) plus 8,000 framing bits.
A collection of 12 frames is called a superframe (SF), corresponding to the format defined in D4, the fourth generation of channel banks, a type of circuit terminating equipment. A collection of 24 frames is used in the extended superframe (ESF) format. T-carrier is a North American standard.
The European and international version is E-carrier. E-1, the fundamental level of E-carrier, supports 30 information channels, each of 64 Kbps. Two additional channels of 64 Kbps are used for synchronization and signaling, respectively, yielding a total 32 of channels requiring aggregate bandwidth of 2.048 Mbps. The Japanese J-carrier standard is similar to T-carrier, but different enough to be incompatible.
Although T-carrier and its offspring were developed to support PCM-encoded voice, they are agnostic with respect to the underlying applications. The transmission facility can't tell the difference between voice bits, data bits, video bits or any other kind of bitsa bit is a bit is a bit. So, digital carrier systems are used in support of the transmission of all varieties and combinations of data in digital format. Further, the traditional channelization conventions often are ignored, as some applications require smaller channels (for example, compressed voice), some require larger channels (for example, high-speed data) and some (for example, frame relay- and broadcast-quality video) don't lend themselves to channelization at all.
There exists a number of techniques for compressing voice in order to reduce the bandwidth requirements. In some cases, there is no perceptible loss in fidelity, while in other cases the loss in quality can be extreme.
Adaptive differential PCM (ADPCM) is a form of waveform encodingone of the main techniques used in speech compression. ADPCM takes advantage of the fact that human speech tends to be modulated in amplitude and frequency in a relatively smooth and gradual manner. ADPCM involves the digital expression (that is, encoding) of only the differences between samples, in the form of a four-bit digital value, which results in a bandwidth requirement of only 32 Kbps. The use of ADPCM enables 48 speech channels, rather than 24, to be carried on the 1.544-Mbps T1 facility.
A considerable number of other compression standards exist, some of which are standardized and others of which are proprietary in nature. For example, the ITU-T G.728 standard defines 16-Kbps audio. Even lower bit rates can be found in the wireless world, where radio channel bandwidth is precious and must be conserved at all costs.
The G.723.1 standard defines 6.3-Kbps audio for use on the "air interface" of cellular and PCS networks. The G.723.1 standard is also an implementation option when voice over Internet Protocol (VoIP) techniques are used. Naturally, there are quality issues at such low-bit rates; nonetheless, the sophisticated coders used in cell phones and VoIP gateways generally do a remarkably good job.
Digital Signal Processors (DSPs)
Special-purpose microprocessors with instruction sets designed to manipulate digital signals are termed "DSPs." Originally developed to support encryption for military applications, DSPs currently are used in a wide range of applications, including modem signal processing, video compression, TV enhancement and straightforward speech compression.
Transmission, Switching and Internet Telephony [return to Table of Contents]
Transmission facilities, also known as transmission systems, comprise the links and circuits that serve to interconnect devices. In the local loop, transmission systems most frequently are in the form of telephone cables made of copper wire. Long-haul systems increasingly are in the form of fiber-optic cables, although orbiting satellites or microwave-radio also are used.
Regardless of the specifics of the transmission systems, there are three primary means of establishing connectivity: dedicated circuits, circuit switching and packet switching. Dedicated circuits are dedicated to the interconnection of two or more devices and are not otherwise shared. Switched circuits allow the network resources to be shared, and often highly shared, thereby eliminating the need for every device to have a dedicated line to every other device. The design of networks takes into account the volume of traffic generated by the users of the network and is usually designed to provide a specified grade of service (GOS), or network availability.
A dedicated circuit is a nonswitched circuit that directly interconnects two or more devices, such as concentrators, switches or routers. Whether analog or digital in nature, the dedicated circuits do not go through any intermediate switching device. Therefore, it is not shared with other devices, attached end users or user groups or applications. Such circuits, therefore, can be rented or leased exclusively to one customer and can be used for voice, data, video, multimedia or virtually any other sort of communications applications. For example, a PBX tie trunk (see the figure "Telephone Network Architecture," above) directly interconnects two or more PBXs via dedicated leased circuits. Companies with high levels of traffic commonly prefer this arrangement, as it can be more economical than the pay-per-call structure of the public long-distance network.
Although a leased line does not go through a central switch, it nonetheless may go through a central office, in which resides the network transmission equipment consisting of wire and fiber optic distribution frames, digital cross-connect equipment, multiplexing equipment and so on.
Circuit switched connections are temporary, continuous and exclusive in nature. That is to say that connections are established between links, or circuits, through a switching matrix on an "as available" basis, and only temporarily (that is, only for the duration of the call). During the course of the call, the connection and all of the associated bandwidth are available continuously and exclusively in support of that call.
In a digital network involving a digital carrier system (for example, T-carrier or E-carrier), each voice call is supported by a specific channel comprising 8,000 specific time slots per second, as defined in the connection setup process. Those time slots are dedicated to the call in progress, even during periods of silence (pauses between words and sentences, staying quiet while the other person talks or other lapses in talking). The channel and its associated time slots remain assigned to that call and are released only when the parties disconnect.
Circuit switching is defined as connection-oriented in nature. That is to say that a physical and logical path is established prior to the commencement of the information transfer, and all data follow the same path during the course of the call.
Packet switching was developed specifically as a cost-effective means of supporting interactive data networking, which is highly bursty in nature. That is to say that the communications are characterized by periods of intense activity interspersed with periods of total silence. Circuit switching is not cost-effective for such communications, and dedicated circuits don't provide the necessary flexibility.
Packet refers to a set of data organized in such a way that it can be accepted, switched and delivered as a unit through an internetwork. Each packet is independent as it works its way through the network, even though it may be part of a stream of packets. In the case of a data transfer involving a large text file or image file, for example, the file is fragmented into multiple payloads of a defined maximum size and each is formed into a packet. The packets are presented to the network in sequence. In the case of voice over a packet network, a number of PCM-encoded voice samples are gathered to make up the payload and formed into a packet. Packets are presented to the network in a sequential stream as long as the voice conversation is active.
In any case, each packet in the sequence works its way through the network independently of others. If some packets are lost, damaged or excessively delayed in transit, measures may be taken to rectify the failure. Alternatively, the failures may simply be accepted, with the specifics of any error control measures being sensitive to the nature of the application being supported.
To facilitate the handling of the packet through the network, various control data is appended. Depending on the specifics of the network protocol or protocol suite involved, the control data is in the form of a header, which precedes the payload, or both a header and a trailer, which trails the payload. In any case, the control data is overhead variously including originating and destination addresses, sequencing, format description, payload type, priority level, path selection and error-control mechanisms.
Developed in the early 1960s in support of the Advanced Research Project Agency Network (ARPANET), X.25 was the first packet data network standard. Although it is fairly primitive by contemporary standards, X.25 remains heavily used. The most popular packet protocol is the Transmission Control Protocol/Internet Protocol (TCP/IP) suite developed in the 1970s for what has become the Internet.
Note: Packet is a generic term for data organized into datagrams and formatted as payloads encapsulated with control information in the form of a header and, perhaps, a trailer. In technology-specific terms, a packet refers to a unit of data (that is, datagram) carried in an internetwork (for example, the Internet), and a frame is a unit of data in a local network (for example, frame relay or LAN). The term cell describes a unit of information carried in a Switched Multimegabit Data Service (SMDS) or Asynchronous Transfer Mode (ATM) network.
Public Switched Telephone Network (PSTN)
The PSTN is a circuit-switched network designed to support analog voice communications. Contemporary PSTNs are largely digital in nature in terms of both transmission and switching systems, although analog local loops from the customer premises to the network edge remain predominant in residential and small-business applications.
Medium and large businesses generally connect to a central office, or end office, at the network edge via digital facilities, commonly in the form of T1 or E-1. Contemporary PSTNs in developed countries use digital facilities to interconnect the switching centers. See the figure "Telephone Network Architecture."
[return to List of Figures]
Telephone Network Architecture
Many facilities and switches are involved in a typical switch train. Designation (4) represents tandem trunks or entire interexchange carrier (IXC) networks. In the U.S., end users pre-subscribe to exactly one IXC, although it can be bypassed in favor of a different IXC on a call-by-call basis through a "dial around" process. Tie trunks (6) are dedicated leased line facilities that bypass the switched network to interconnect Private Branch Exchanges (PBXs) in a multisite enterprise.
A traditional PBX is a privately owned circuit switch located on the customer premises. A PBX enables people within the enterprise to talk to each other (station to station), to gain access to the public network for purposes of placing outgoing calls and to receive incoming calls from the public network. The latter two are accomplished through the previously mentioned analog or digital trunk facilities.
Note: A trunk is defined as a link that interconnects switches. A PBX trunk interconnects a PBX switch and a PSTN central office switch. A tandem switch is a PSTN switch that interconnects other PSTN switches, such as central office switches.
Internet telephony refers to communications servicesvoice, facsimile and voice-messaging applicationsthat are transported over the Internet, a packet network based on the TCP/IP protocol suite. In general, use of the Internet in this manner bypasses significant portions, but not all, of the PSTN. See the figure "Voice on IP Networks."
[return to List of Figures]
Voice on IP Networks
The top half of the figure shows a voice call routed to the Internet. The lower half shows a company's private internal network (intranet) carrying voice calls and data across an IP-based wide area network (WAN), as an example of voice/data convergence.
Phone to Phone on the Internet
The five steps involved in an Internet telephony call, performed by the originating network and reversed at the destination, are listed in the table "Steps in an Internet Telephony Call." Also listed are the entities at the respective ends of the transmission facilities as the call progresses. For simplicity, calling and called parties are assumed to be using analog telephones.
[return to List of Tables]
|Description||Entity at One End||Entity at Other End|
| Call origination in the normal manner from an analog telephone||Calling party||PSTN originating end office, which routes the call to the originating Internet Service Provider (ISP)|
| Conversion of analog voice signal to digital (PCM) format||PSTN originating end office|||
| Local transmission of PCM voice signal on interoffice trunk (T1 or E-1)||PSTN originating end office||Originating ISP's gateway equipment|
| Compression/translation of PCM voice signal into Internet Protocol (IP) packets||Originating ISP's gateway equipment|||
| Transmission of IP voice signal on the Internet||Originating ISP's IP packet-switching infrastructure (packet switch, router)||Internet backbone|
| Transmission of IP voice signal on the Internet||Internet backbone||Destination ISP's IP packet-switching infrastructure|
| Decompression/translation of IP voice signal into digital (PCM) format||Destination ISP's gateway equipment|||
| Local transmission of PCM voice signal on interoffice trunk (T1 or E1)||Destination ISP's gateway equipment||PSTN terminating end office, which routes the call to the called party|
| Conversion of PCM voice signal to analog format||PSTN terminating end office|||
| Call delivery in normal manner to an analog telephone||PSTN terminating end office||Called party|
PBX to PBX on an Intranet
In a traditional PBX environment (lower half of the figure "Voice on IP Networks"), employees place IP telephony calls using the same steps (in the table "Steps in an Internet Telephony Call"), but are preceded by a dial access code (usually "9" in the U.S.) to reach an outside line. Should the PBX be IP-enabled, it will perform the necessary gateway functions (that is, protocol conversion) to provide compatibility with legacy end-office circuit switches over traditional analog or digital channelized circuits.
Terms such as Internet telephony, IP telephony and VoIP are often used, or misused, interchangeably to denote similar types of calls and services. But Internet telephony refers to calls that are transported specifically over the Internet, while IP telephony typically refers to calls that are carried on any public or private network running the IP protocol. And VoIP is more properly a technology banner, implying defined (open) implementation standards. By convention, the public Internet is capitalized, while private intranets are not, even though both are IP-based and intranets generally run over the Internet.
Packet Telephony Pitfall
As described above, circuit switching assigns time slots and dedicates them to the switch train, guaranteeing that talk paths are there when someone talks.
This very obvious and comforting situation is compromised when speech is chunked into packets and sent into a network designed to carry data packetsnot digitized speech packetsand switched onto different paths to or toward the destination. Time slots are not fixed along the same physical path for the duration of the call; indeed, it is possible that each packet could take a different physical path. As the length of the physical path has a direct impact on the length of time that it takes a signal to travel between two points, the result of multiple physical paths is jitter, which is variability in packet latency, or delay. Delay and jitter also can be the result of variations in network congestion levels from packet to packet.
As packet data communications are bursty in nature and as a great many transmissions commonly are supported over the Internet at any given moment, congestion levels are variable and unpredictable. As congestion increases, packets may be held in buffer space in switches and routers at various points in the network until the congestion eases and they can be forwarded. If a given packet is lost or errored in transit, it typically must be retransmitted. Variations in physical paths, congestions levels, and packet loss and error all contribute to latency, jitter and loss of sequencing.
Typical data communications applications, such as e-mail and file transfers, can deal with these issues quite effectively, as there is always time to recover from latency, jitter and loss of sequencing in order to reconstitute the data in its original form. Stream-oriented communications, such as real-time uncompressed voice and video, have a difficult time dealing with these issues, however, as time is of the essence and data must be played immediately and in the order in which it is received.
Intelligent Networks (INs) [return to Table of Contents]
The concept of INsbased on digital switching, database intelligence and ITU-T SS7 (Signaling System 7) signalingprovides customers with a host of advanced network features and services, such as distributed call processing, one-number services, class of service (CoS), private network services and network billing enhancements, to name a few. As shown in the figure "Basic Intelligent Network (IN) Including SS7 Signaling," the elements of an IN are:
[return to List of Figures]
Basic Intelligent Network (IN) Including SS7 Signaling
SS7 signaling links connect SS7 network nodes. SSP voice paths (to PBXs, subscriber phones and other SSPs) are omitted for clarity.
SSPs are typically end offices, but can be tandem switches in sparsely populated areas with capabilities that go beyond straightforward switchingenhanced software and hardware enable them to communicate with application databases. The SSP formats and sends a request (an SS7 message) to an SCP, where an application database is stored, and suspends call progress until a response is received that provides the routing information needed to complete the call. An SSP may communicate with many different SCPs, depending on the number and variety of applications available.
STPs are essentially packet-switching systems used to transport messagescall setup messages and routing request messagesbetween SS7 nodes. As a cost-effective alternative to interconnecting all nodes directly to one another, STPs serve as centralized hubs in the SS7 network. Many nodes are linked to a single STP, and in turn all STPs are interconnected. Messages sent to an STP are routed to the correct destination node. Because an STP connects to many SS7 nodes, it must be capable of handling high-message throughput and must be equipped to support many signaling links. Redundancy and reliability are also key attributes.
SCPs are the centers of intelligence in the network. The main function of an SCP is to accept a query for information, retrieve the relevant information from the appropriate database and send a response to the originator (typically an SSP) of the request. Adding or updating databases, without affecting any other node in the network, can increase the functionality of an SCP.
Integrated Services Digital Network (ISDN)
The ISDNdefined and described in ITU-T Recommendationsextends to the customer premises the capabilities and benefits of the IN to support a wide range of voice and nonvoice applications over the same digital network. ISDN is an end-to-end (customer premises-to-customer premises) digital network that integrates enhanced voice and image features with high-speed data and text transfer.
ISDN provides two interfacesbasic rate interface (BRI), also known as basic rate access (BRA), and primary rate interface (PRI), also known as primary rate access (PRA). Both interfaces consist of B channels and D channels. The B channels, or bearer channels, are information-bearing channels that provide transparent digital channels for voice or high-speed data transmission at 64 Kbps per channel. The D channel, or delta channel, provides a nontransparent channel for signaling, telemetry and low-speed packet data at 16 Kbps or 64 Kbps:
Mobile Communications [return to Table of Contents]
Modern mobile communications revolve around two main ideas. The first, "smaller is better," breaks a relatively large geographic area into many contiguous cells. So a single macrocell is broken into many microcells, which can be further subdivided into many picocells. Radio towers situated approximately at the center of cells communicate with mobile telephones (cell phones) and other devices (such as mobile personal digital assistants, or PDAs).
The second concept, "less is better," reduces the power transmitted from respective towers, which is the basic reason that the cells can be reduced in size. Taken together, reduced transmitted power and small cellular areas enable radio frequencies to be reused. This has a multiplication effect, enabling many more users to be served using the same range of frequencies.
Cellular/Mobile Wireless Basics
The centerpiece of cellular/wireless networks is the mobile switching center (MSC), which interconnects small radio coverage areas into a larger system. See the figure "Mobile Communications Network."
[return to List of Figures]
Mobile Communications Network
Base stations communicate by radio with mobile phones or other wireless devices. Land lines connect base stations to an MSC, which tracks the location of devices that have been turned on.
When a device such as a mobile phone is turned on, it scans for an unused command channel, locks on and sends a registration request. To originate a call:
To receive calls (known as "call delivery"), the mobile phone must be turned on and locked to a command channel. Assuming the call originates from a residence:
Handoff to the Same MSC
When a mobile phone is turned on ("available") or is turned on and engaged in a conversation ("busy"), it may traverse a cell site and find itself in a different cell site. At some point, the received signal strength of the command channel or the talk channel will decrease to the point where reception is no longer viable. The mobile phone monitors its received signal and sends signal strength measurements to the MSC. Signal strength below a predetermined level triggers handoff (of the mobile phone) to a base station with a stronger signal. A change in frequency occurs, coordinated by the MSC, as adjacent cell sites operate on mutually exclusive frequency "lists."
Some cellular networks make "hard handoffs" through a "break and make" process that involves breaking the connection with the original cell prior to making the connection with the next. This hard handoff is imperceptible in voice communications, but may cause a data session to be terminated. Other networks make "soft handoffs," which are acceptable in data communications as the connection is made to the next site before the connection is broken at the original site.
Handoff to a Different MSC
Conceptually, the situation is similar when the new cell site is served by a different MSC. The details are complex, though, as the MSCs must exchange messages (IS-41 or SS7 messages depending on the network) to coordinate many actions:
After successful handoff, the talk path consists of PSTN switches and facilities, a trunk to the first MSC and the first MSC's switch, a trunk to the second MSC and the second MSC's switch, a trunk to the new base station, a radio channel to the mobile phone and the mobile phone itself. Theoretically, the process can continue indefinitely, the talk path accumulating additional MSCs (and trunks) during the mobile phone's apparent odyssey. Practically, calls are finite, and even the longest ones terminate before the switch train grows beyond several MSCs.
Users sometimes transport mobile phones out of the home service area, a condition known as "roaming." When the phone is turned on, emitting its mobile phone number, the MSC serving the mobile phone at the distant location communicates with the home location register of the subscriber's service provider to find out whether or not to provide service to the visiting phone (credit worthiness), as well as the service options and phone features that it should (or should not) honor. In this way, the service provider knows the whereabouts of the mobile phone, specifically, it knows the identity of the visited MSC, and the visited MSC knows how to provide service as if the subscriber resided there. Call origination is handled by the visited MSC and is the same as described above (that is, the visited MSC treats the roaming phone as if it were one of its own). Call reception, however, is complicated by the fact that the phone is roaming.
Assume the call comes from a residence in Chicago, for example. It will be directed, via the PSTN, to the home location register of the subscriber's service provider (say, Dallas), which will notice that the mobile phone is out of the home service area. The following steps now occur:
TLDNs are not published, as they are reserved for internal network routing only. During a random call origination, if a TLDN is dialed in error, the PSTN will route the call to the MSC (assumed to be in New Orleans), which will provide an announcement that the call cannot be completed as dialed.
Wireless Communication Standards
There are a number of different mobile radio systems ranging from pagers to the pan-European digital cellular radio system known as Global System for Mobile Communications (GSM). There also exists an array of mobile communications technologies that focus on a variety of sub-segments of the mobile communications marketplace. That marketplace can be further subdivided into segments that address different aspects, characteristics or needs of its customer base.
There exist a great number of mobile phone systems and standards, all of which are incompatible. The narrowband cellular voice systems currently in use span 1G (first generation) analog and 2G (second generation) digital systems, the most significant of which include:
Data communications is difficult in 1G and 2G systems. The only datacom technology that has experienced any real success is Cellular Digital Packet Data (CDPD) in the U.S. Based on TCP/IP, CDPD is a packet data communications technique that operates over established AMPS analog networks at theoretical rates up to 19.2 Kbps, which generally translates into a practical data rate of no more than 9.6 Kbps.
Emerging 2.5G and 3G cellular networks all fit under the umbrella of International Mobile Telecommunications-2000 (IMT-2000), an initiative of the ITU-T intended as an international digital wireless network architecture for the twenty-first century. The various 3G specifications include the following speeds and intended applications:
Note: The above data rates all are theoretical best case rates. Actual data rates usually are much lower due to factors such as ElectroMagnetic Interference (EMI), Radio Frequency Interference (RFI), signal attenuation and line-of-sight issues.
A shortlist of next-generation 2.5G and 3G wireless standards, and their theoretical maximum data rates, includes the following:
Clearly, the confusion over cellular standards is likely to continue for some time. Multimode terminal equipment resolves these issues of incompatibility in some cases. There are a number of other voice and data mobile wireless solutions, some of which are standard and others of which are proprietary in nature.
Wide-Area Data Networks [return to Table of Contents]
Broadband ISDN (B-ISDN)
Early in the development of ISDN, the necessity was recognized for an advanced form capable of carrying multimedia information at rates of hundreds of millions of bits per second. In the ITU, two main service categories have been defined: interactive services and distributive services.
Interactive services are subdivided into conversational services, message services and retrieval services. Conversational services are usually bidirectional, although in some circumstances they can be unidirectional and in real time between users, or between a user and a host. Examples include videoconferencing and high-speed data transmission. Message services will offer communication via storage units, such as a mailbox or as message-handling functions, which include not only speech but also moving pictures and high-resolution images. Retrieval services offer user access to information stored centrally and accessed on demand. Examples of these services include film and high-resolution images, together with audio.
Distributive services are differentiated between those services with user presentation controlsuch as broadcast services for TV and radioand those with individual user control. The availability of high bandwidth enables a number of different types of information to be supported by one service, resulting in the development of multimedia services. For example, video telephony includes audio and video and possibly text and graphics. Many of the broadband servicessuch as video signal transmissionrequire variable bandwidth, which is best met by a packet-based technology. For this reason, ITU chose ATM as the target transfer mode for B-ISDN.
It is important to note that ATM is a transfer mechanism and as such is, in principle, independent of transmission technology. It is a fast-packet switching technique that uses short, fixed-length packets called "cells." As such, ATM is also referred to as a "cell-switching technology."
In principle, it is quite similar to other packet-switched techniques; however, the detail of its operation is somewhat different. Each ATM cell is made up of 53 octets, of which 48 octets generally make up the user information field (payload) and five octets generally make up the header (overhead). The header identifies the "virtual path" to be used in routing a cell through the network. The virtual path defines the connections through which the cell is routed to reach its destination. ATM is a form of TDM. It differs from synchronous multiplexing in that channel separation is not dependent on reference to a clock.
Frame relay also is a form of fast packet switching. Actually an ISDN framing convention intended for LAN interconnection, frame relay is a technique for gaining access to a packet data network, such as TCP/IP or ATM. It is used primarily in data communications environments, although standards-based voice over frame relay (VoFR) implementations are not uncommon.
As an ISDN spin-off and an interim technology designed primarily to serve both local-area network (LAN) interconnection and host computer environments, frame relay achieves about 10 times the packet throughput of X.25 packet-switching networks by letting information move across a network guided and checked by the following seven core functions of Link Access Procedure on the D-channel (LAPD):
A frame relay limitation is its lack of a mechanism for error detection and correction. These functions are left to the end nodes in the network. This is a major QOS issue when considering frame relay for the transport of voice communications. This lack of error control is not an issue in most data communications applications, as there generally is time to recover from data loss or error. As a point of fact, the fully digital nature of the transmission facilities, which generally are largely optical, and switching elements results in relatively little loss or error.
Frame Relay, X.25 and TCP/IP Compared
Unlike its predecessor X.25, frame relay does not store and forward data frames, but rather simply switches to the destination part way through the frame, thereby reducing transmission delay considerably. Because storage requirements are minimal, frame relay is more cost-effective than X.25.
X.25 and TCP/IP are similar in that they are both packet-switched protocols. However, they differ in a number of areas:
There are a number of specific technologies in the generic Digital Subscriber Line (xDSL) family. Asymmetrical Digital Subscriber Line (ADSL), the consumer-level service with which most users are familiar, is a broadband service currently being offered by local exchange carriers (LECs) and ISPs in many areas where the local loops can support the demands of the technology.
ADSL delivers broadband Internet access over UTP local loops up to 18,000 feet in length, as long as the loops meet fairly stringent requirements. For example, load coils are not acceptable, and mixed gauges and bridge taps are undesirable. For technical reasons that are beyond the scope of this article, ADSL runs in asymmetric mode, with considerably more bandwidth provided downstream (that is, from the network edge to the customer premises) than upstream. Currently, the maximum rates offered by most carriers are 1.544 Mbps downstream and 256 Kbps upstream.
The standards describe considerably higher rates, although the higher frequency signals demand that the distances be shorter and that the overall quality of the local loops be fairly exceptional. A great advantage of ADSL is that the digital packet data channel is "always on," meaning that it is not necessary to dial through the PSTN, as is the case with traditional dial-up modems. This not only eliminates a step, but also eliminates both dial-up delays and busy signals. Additionally, ADSL offers the advantage of simultaneously supporting an analog channel for voice and fax communications.
In addition to ADSL, the xDSL family comprises Very High Data Rate DSL (VDSL), High Bit Rate DSL (HDSL), Single Line DSL (SDSL) and ISDN DSL (IDSL), among others.
Cable modems recently have emerged to compete with ADSL in the broadband access market. Cable modems are customer premises equipment (CPE) access devices that enable computer equipment to connect to the Internet over a digital cable television (CATV) network that simultaneously supports a great number of entertainment TV channels. In the most prevalent scenario, a PC is connected to an external stand-alone cable modem via an RJ45 Ethernet network interface. The cable modem is attached to a CATV network via an F connector, which is a 75-ohm coaxial cable connector commonly found on televisions and videocassette recorders (VCRs).
PCs are not the only equipment that can be connected to cable modems. As small office/home office (SOHO) environments expand and cable access extends into the business environment, cable modems can be connected to hubs, switches and routers to allow networks access to the Internet. Some cable modems are now including routing and four-port hub capabilities into a single cable modem device. Newer versions of cable modems include USB connections for PC connectivity and peripheral component interconnect (PCI) cable modem cards.
While the bandwidth that cable modems can support is impressive, it is shared among all users on the same CATV network segment. Performance will vary depending on the number of users who are online simultaneously and the type of work each person is doing. Access speeds can and generally will be significantly less than what is theoretically possible. While bandwidth can be supported in the 10-Mbps range and higher, more realistic bandwidth numbers are in the 1-Mbps to 3-Mbps range for downstream traffic, and 250 Kbps to 2.5 Mbps for upstream traffic.
Synchronous Optical Network (SONET) and Synchronous Digital Hierarchy (SDH)
Digital telecommunications networks in North America (T-carrier), Europe (E-carrier) and Japan (J-carrier) were not based on a common standard, a fact which makes their internetworking somewhat cumbersome. Issues of compatibility aside, these conventional digital transmission networks simply didn't have the capacity to handle increasing traffic demands. Fiber optics certainly had the bandwidth to meet those demands, but all of the established systems were proprietary in nature, which meant that they could not interconnect, and therefore the carriers were not able to enjoy the benefits of deploying multivendor networks.
At the request of its client/owners, Bellcore (now Telcordia Technologies) proposed to American National Standards Institute (ANSI) the development of the. SONET standards for optical transmission in 1985. In 1988, the first phase of SONET standards were released. The Comite Consultatif International Telegraphique et Telephonique (CCITT), which is now the ITU-T, was following SONET progress and began work on an international variation in 1986. The World CCITT SDH standards in its 1988 Blue Book. Although SONET and SDH differ, primarily at the lower multiplexing levels, those issues are fairly readily resolved in situations where interconnection is desirable. SONET/SDH and ATM are now the core transport and switching technologies, offering network operators and end users several advantages, including:
SONET/SDH is built on T-carrier standards. Essentially, SONET is a high-speed optical version of T3, which is a T-carrier multiplexing level supporting the equivalent of 28 T1s and running at a signaling rate of 44.736 Mbps. SONET adds some overhead for network management purposes, which brings the signaling rate up to 51.84 Mbps for Optical Carrier-level 1 (OC-1), which is the foundation level. Like T-carrier, SONET is TDM-based, with frames transmitted 8,000 times per second at the precise pace of 125 microseconds. The frame structure is nine rows by 90 columns, each of which columns is a byte wide. Of the total 810 bytes, 765 bytes are payload, organized as nine rows by 85 columns.
Should OC-1 not be satisfactory, the next step in sequence typically is OC-3, which comprises three OC-1 frames and which runs at a signaling speed of 155.52 Mbps. OC-3 is the foundation level for SDH. At that rate, all three OC-1 frames are transmitted in sequence every 125 microseconds. The SONET/SDH hierarchy currently includes OC-768, which is the equivalent of 768 OC-1 frames multiplexed and running at a signaling speed of 39.813 Gbps.
On analog public telephone networks, transmission facilities can be two-wire or four-wire circuits. A two-wire circuit consists of two copper wires, each in a color-coded, insulated covering; the two wires are loosely twisted around one another to improve the performance characteristics of the circuit. At the most fundamental level, a four-wire circuit is a pair of two-wire circuits, with each pair supporting transmission in one direction. Twisted-pair wire comes in many forms: some cables are waterproof, some have fire retardant coverings and some are shielded for extra protection against electrical interference.
In a local environment, in which terminals (for example, client workstations and printers) are attached to a server in proximity, data transfer generally occurs over data grade unshielded twisted pair (UTP) wiring. Over distances of 100 meters or so, data-grade UTP performs well and at much lower cost than coaxial cable.
Coaxial cable, which was heavily used prior to the introduction of UTP in LAN applications, consists of two conductors. A solid core center conductor supports data transmission, while an outer conductor of solid metal foil or metal mesh acts as a shield against ambient sources of electromagnetic interference. A "dielectric" (that is, a material that does not conduct direct electric current) separates the two conductors. A cable that consists of two central conductors in the same mesh tube is called "twinaxial cable." As the gauge of the center conductor in a coaxial cable is relatively thick, it offers relatively little resistance to an electrical signal and therefore supports much higher frequency signals than does twisted pair. That translates into greater bandwidth over longer distances, which explains its traditional use in both CATV networks and data centers. However, coaxial cable is expensive, bulky and inflexible.
The increase in availability and lower prices of fiber optic cabling and equipment has resulted in its incorporation into a large proportion of new networks. This is especially the case where there is a need to future-proof installations against the rising demands for bandwidth. As discussed in some detail above, SONET/SDH resolved concerns about the lack of standards in the fiber optic domain. More recently, optical multiplexing techniques have developed in the form of Wavelength Division Multiplexing (WDM), which allows multiple wavelengths to share the same optical fiber, much as multiple electrical signals can share the same copper cable through Frequency Division Multiplexing (FDM).
[return to List of Figures]
The three basic types of communications cabling: twisted pair (in this example, shown in a 25-pair bundle), coaxial cable (coax) and fiber optic cable.
Microwave communications describes point-to-point terrestrial radio systems operating within the frequency range of 1GHz to 30GHz. Once exclusively the domain of the common carriers, microwave communications has become a major competitor of standard wireline telephone communications. Generally speaking, microwave communications involves sending information via high-frequency radio between a transmitter and a receiver. As such, high-frequency signals require a clear "line of sight"; there must be no obstructions such as buildings or trees in the path between transmitter and receiver. Contemporary microwave systems are digital in nature and offer signal quality to copper-based conducted transmission media.
Satellite networks are essentially nonterrestrial microwave systems with signal repeaters placed in space. Satellite transponders, which are the operational elements of satellites, receive radio signals originating from an earth station (that is, terrestrial antenna), filter the noise out of the signal and amplify it, shift the frequency to avoid interference between incoming and outgoing signals, and transmit them back across the assigned footprint (that is, coverage area), where any number of terrestrial antennas can receive it simultaneously.
Most contemporary communications satellites are Geosynchronous Earth Orbiting (GEOs) operating in equatorial orbital slots at altitudes of approximately 22,300 miles. Once positioned properly, the satellites rotate around the earth at the same speed as the earth rotates on its axis. In other words, they are synchronized with the earth (geo). Also known as geostationary, the satellites are therefore always in the same position relative to any point on the earth's surface.
Data Integrity: Error Detection and Correction
Any communications circuit or system can suffer from interference. Electrical storms, electromagnetic fields from electric motors, cross-talk from other circuits in proximity and other phenomena can impact the circuit or system and introduce errors into the transmission. Twisted-pair copper cable systems and radio systems are particularly susceptible, but they all can be affected. Whether it's one bit or several thousand bits that are errored or lost in transit, the integrity of the information is compromised. Every errored or missing bit is a potential catastrophe that must be identified, isolated, diagnosed and corrected in some way, or not. Some applications just don't have the time to deal with error and loss. Rather, they just accept it. Real-time uncompressed voice and video are examples.
Those exceptions aside, a typical file transfer or program download might involve thousands or millions of bytes. Many thousands of such files might transit the network at any given time at speeds of thousands or millions or even billions of bits per second, and the loss of even one bit could alter a character or control code. Therefore, data communications generally requires stringent accuracy controls. These controls consist of bits added to characters and blocks of characters at the sending end of the line. These bits are then checked and verified at the receiving end of the line to determine whether data was lost or errored during transmission. Two basic data communications controls include:
In data communications, parity denotes a relatively simple error-checking technique in which the 1 bits in a character are summed and a 1 or 0 parity bit is appended to create either an odd value (if odd parity is desired) or an even value (if even parity is chosen) prior to transmission. The receiving device examines each received character for parity. If the devices are set for odd parity and a received character has an even parity, it is assumed that one or more bits were corrupted and that the character is in error. The situation is similar in an even parity system if the receiver calculates an odd bit value.
The fundamental parity checking technique is known as vertical redundancy checking (VRC), in reference to the fact that human beings add numbers in vertical columns and the process is redundant, with the receiving device repeating the mathematical process executed by the transmitting device. VRC is highly unreliable, as an even number of errored bits creates an undetectable character error. The reliability of the technique is improved considerably if longitudinal redundancy check (LRC) is also used. This combined approach involves the process of checking for errors in a block of characters by examining parity values both vertically and horizontally as though the bits were in a matrix format. In fact, it would take an odd set of compensating bit errors to create an undetectable error in a block. Parity checking a fairly primitive technique used in asynchronous communications.
CRC, a more powerful error-detection technique used in synchronous transmissions, uses calculated polynomial check characters. CRC views an entire block of data as one long binary number that is divided by another fixed binary number. The result of the calculation is a summary description of the data, which is truncated, generally as a 16- or 32-bit value. This is expressed as eight-bit block-check characters (BCCs) that are appended to the data block prior to its transmission. BCCs appear in the Frame Check Sequence field of HDLC frames (see the previous figure, "HDLC Framing"), for example. At the receiving end, the process is repeated as the BCCs are recomputed from the received data and compared to those appended to the block by the transmitting device. If the BCCs agree, the block is declared error-free. If they disagree, it means that one or more errors exist in the block. The entire block is suspect and must be retransmitted.
Since an error detected in a message must be corrected fairly immediately, a transmitting device must receive an acknowledgment on a near real-time basis of the received data's accuracy. Because data must be retransmitted if there is an error during transport, transmitting equipment must store all information that has not been acknowledged by the receiving station.
There are a number of error-correction techniques, and more are being developed all the time. The two basic methods are "Stop and Wait" Automatic Repeat Request (ARQ) and "Go back N" ARQ:
Yet another alternative is that of forward error correction (FEC). This technique involves the embedding of redundant characters in the data block. This information allows the receiving device to identify, diagnose, isolate and correct errors without the necessity for a retransmission. FEC is widely used in networks where bandwidth is limited and expensive, with cellular data communications being but one example.
Local Area Networks [return to Table of Contents]
A LAN is a communications network that is usually owned and operated by the enterprise customer. A LAN operates over a limited geographical area and enables many independent peripheral devices, such as PCs and terminals, to be linked to a network through which they can share centralized hosts, files, applications, printers and various other resources. Some organizations (for example, banks or financial institutions) have enough data traffic within a city to make intracity networking viable. In this case, individual LANs may be interconnected to form a metropolitan area network (MAN).
LANs are generally fault-tolerant, incorporating a simple architecture with control distributed among participating stations. Since the entire network does not depend on a single polling or switching device, the failure of one component does not bring down the entire LAN. Standard LAN protocols largely emanate from the IEEE 802 working group. Two major LAN standards are:
Ethernet was developed by Xerox and subsequently modified and standardized by the IEEE as 802.3. That original Ethernet ran at 10 Mbps over copper coaxial cable. In contemporary usage, the preferred medium is data grade UTP wiring, although STP and coaxial cable sometimes are used.
As packet networks sharing a common medium, LANs must make use of some sort of access mechanism in order to manage contention for these limited resources. The access technique use by Ethernet is carrier-sense multiple access with collision detection (CSMA/CD). This technique requires that each workstation listen to the network (that is, sense the carrier frequency) to determine if it is available before transmitting. If, however, two or more workstations are transmitting data over the same segment simultaneously, their signals will collide and the data will be destroyed. CSMA/CD allows the workstations to detect this collision, at which point each backs off and calculates a random time interval, after which it attempts to access the network and retransmit the data.
CSMA with collision avoidance (CSMA/CA) offers an alternative technique that is particularly useful in wireless LANs (WLANs). With CSMA/CA, a device senses the carrier and issues a request to send (RTS). If the channel is available, the destination device issues a clear to send (CTS), which essentially advises the other devices to back off and allow the communication to take place without contention.
As the volume of network traffic increased, however, the bandwidth offered by traditional 10-Mbps Ethernet LANs became inadequate in many applications. The first step in alleviating these bottlenecks was the introduction of 100Base-T, which makes use of relatively simple hubs to which terminal connect via Category 5 (Cat 5) data grade UTP at speeds up to 100 Mbps. While essentially a brute force attack on congestion, 100Base-T is effective. The next step was the development of switched Ethernet, which provides multiple 10/100-Mbps paths for a limited number of devices through the switching matrix, rather than a single 10/100-Mbps "pool" of bandwidth for which all devices must contend.
More recently, Gigabit Ethernet (GbE) surfaced. Generally implemented as a switched Ethernet solution, 1000Base-T originally required the use of fiber optic cabling, although high performance Cat 5 UTP can perform acceptably well over distances up to 100 meters if installed correctly. Cat 6 cables, which were introduced in 2002, are designed specifically to support 1000Base-T.
Developed by IBM, Token Ring typically runs on a ring topology, although star networks based on switches are also possible. There are never any collisions in Token Ring. The first station to switch on in a network owns the "token," a unique sequence of control bits. While it has this token, it is capable of transmitting data, and no other workstation can take the token on that revolution. After the current owner completes transmitting its data, it passes the token to the next workstation on the network, and so on.
There are three basic LAN topologies: linear bus, ring and star. The topology can be defined as the physical layout of the network.
[return to List of Figures]
The Bus topology is typical of Ethernet networks; rings are associated with token-passing (Token Ring) networks.
In linear bus topology, stations are arranged along a single length of cable that can be extended at one of the ends. A tree is a complex linear bus in which the cable branches at both ends, but which offers only one transmission path between any two stations. All broadband networks and many baseband networks use a bus or tree topology.
In a ring topology, stations are arranged along the transmission path so that a signal passes through one station at a time before returning to its originating station; the stations form a closed circle.
A star topology has a central node that connects to each station by a single point-to-point link. Any communication between one station and another must pass through the central node.
In bus and ring networks, all transmissions are broadcast. Any signal transmitted on the network passes all the network's stations. In star networks, signals sent through the central node are switched to the proper receiving station over a dedicated physical path.
The LAN market can be divided into two distinct segments: large-scale LANs and server-based LANs:
A LAN centralizes the control of an organization's distributed computing resources and ensures that each department's PCs are compatible with the network and with machines from other departments. Ideally, through a LAN the manager can make sure that all company decisions are based on the same data. When used properly, the LAN provides a common interface for a diversity of otherwise incompatible equipment.
In the office, the LAN can give users fast and efficient access to a common pool of information, such as customer lists, schedules and document formats. It also allows an entire office to share expensive resources, such as printers and duplicators, thereby streamlining the production and distribution of paper documents. Ultimately, a LAN can eliminate the need to circulate paper documents by electronical distribution (e-mail, instant messaging) of memos and other textual materials. In an automated factory or laboratory, the LAN can simplify the process of "retooling" by allowing the user to download data from a number of programmable devices simultaneously from a central site. It can also isolate failures and bottlenecks in plant operations.
At one point, interest in fiber optic LANs was increasing rapidly, with several fiber optic systems available for LAN applications based on a mature optoelectronic technology. More recently, however, interest in this technology waned as copper cabling became more sophisticated, with Category 5 (Cat 5) easily supporting data rates of 100 Mbps, and even 1 Gbps. Optical fiber remains highly viable over distances greater than 100 meters, or where electromagnetic interference or security is an issue.
The Internet now has millions of hosts connecting many millions of people all over the world. Educators, consumers, telecommuters, librarians, hobbyists, researchers, government officials and businesses are among the groups today that use the Internet for a variety of purposesfrom communicating and collaborating with one another to accessing valuable information and resources.
The Internet provides connectivity for a wide range of application processes called "network services." For example, users can exchange electronic mail, instantly message each other in groups, access and participate in discussion forums, search databases, browse indexes and transfer files. Also, use of the Internet for multimedia applications, including voice, is still in the early stage.
Internetworking refers to the connecting together of two or more networks, which may be LANs, WANs or a mixture of the two. As the Internet continues to grow in size, popularity and efficiency, as LANs proliferate in business environments and as enterprises rely on several communications networks simultaneously, managers seek better ways to move information from one network to another. Internetworking devices take a LAN signal and send it further than the original LAN specification allows.
Devices generally fall into three types, thoughas so often happens in communicationsa single product can incorporate more than one function:
[return to List of Figures]
Bridges, Routers and Gateways
Bridges operate at the data link layer of the OSI model, routers are the network layer devices that select pathways to send data to destinations and gateways operate at the upper layers of the OSI model.
Communications Equipment [return to Table of Contents]
Private Branch Exchanges (PBXs)
A PBX is a telephone switch located on an enterprise's premises that primarily establishes voice-grade circuits over access lines between individual users and the PSTN. The transmission of wide area PBX calls is still typically over the PSTN. However, enterprise telephony calls are increasingly being routed through gateway devices over Ethernet-based LANs and WANs, the IP-based Internet, and even ATM and frame relay networks.
A PBX is a private system in that it is typically used by one organization or one building complex with capacity requirements ranging from less than 100 lines (but typically above 40 lines) up to thousands of lines. Branch suggests a remote subsystem, as the first PBXs were like small partitions of central office switches located on a customer's premises. In telephony, an exchange is defined as a device that controls the connection of incoming and outgoing callsin short, a switch.
In addition to the solid reliability and performance that have been characteristics of traditional PBX technology over its life cycle, advanced functionality, such as IP telephony, call center technology, electronic messaging (voice, fax and even unified messaging), computer-telephony integration (CTI), broadband capabilities, PBX networking and in-building wireless communications, have all been introduced as PBX enhancements over the past decade or so. PBXs also have become less proprietary as many now feature open application programming interfaces (APIs) that encourage third-party software development.
In traditional digital PBX systems today, time division multiplexing (TDM) is still the most commonly deployed switching matrix design. However, the packet-switching designs in newer IP-based telephony systems, in particular, are rapidly pushing TDM and circuit switching to the background. Traditional circuit-switched PBXs are now supporting interfaces to IP, ATM or frame relay infrastructures and equipment via an assortment of new software, gateways, trunk cards and line cards. For example, most PBXs today support VoIP functionality in various ways, ranging from offering VoIP gateway interfaces to IP networks, with supporting IP line/trunk cards, and IP telephones, or by adding VoIP functionality to its core switching matrix.
Key Telephone Systems (KTS)
Smaller organizations (typically with less than 40 stations) use a KTS rather than a PBX. Other than size, a major difference between a PBX and KTS is that a key system does not utilize an operator console. The heart of a KTS is a key service unit (KSU) that is the common control cabinet for all major system operations and functions. A KSU performs CO line connections, intercom functions, paging and station connections. Each KTS extension has a lamp indicator for all available outside lines, showing whether or not they are busy as well as giving visual indication of an incoming call. A KTS does not require dialing a code to gain access to outgoing lines; PBXs invariably require dialing a number or code (such as "9"). A hybrid system typically provides the combined features and benefits of both KTS and PBX systems.
Automatic Call Distribution (ACD)
ACD is an important application for businesses handling large volumes of incoming telephone calls, such as in a call center environment. A PBX system equipped with integrated ACD software enables the switch to automatically route incoming customer calls to groups of call center agents with specific skill sets, agents who have been idle the longest or agents who have handled the fewest number of calls.
In many large call center applications, PBXs interface with stand-alone ACD systems. Leading PBX/ACD vendors also offer integrated ACD solutions for their PBX systems, giving customers a choice between implementing a PBX/ACD solution with better integration capabilities or a third-party solution from a stand-alone ACD vendor.
The agent capacities of PBX/ACD systems allow those vendors to compete for large call center installations, traditionally the domain of stand-alone ACD vendors. In addition, the trend toward networked call centers has advantages for the PBX/ACD vendors, who see themselves as having deeper skill sets and more expertise in network communications than their stand-alone ACD counterparts.
Today's client/server-based ACD solutions off-load many of the call-processing tasks that require application-based intelligence to standard PC-server platforms. Utilizing CTI technology and API software, these solutions function as call/connection servers within the enterprise network. Call center businesses of all sizes have substantial investments in modern ACD call center technology. As such, ACD remains a major revenue stream for PBX vendors.
With the shift toward supporting e-commerce and customer relationship management (CRM) solutions and applications, along with the increasing emphasis on handling all contact mediavoice calls, faxes, e-mails and Web-based inquiriesACD vendors and the market, in general, now promote use of the term "contact center" instead of "call center." Attributes of a contact center include:
Voice Processing (Mail) Systems
A voice mail system records, stores and plays voice messages. It supports features that enable end users to access, forward, reply to, schedule the delivery of and tag/edit messages, among other functions. End users, or subscribers, are the owners of personal voice mailboxes in a voice mail system. Subscribers can access messages in their voice mailboxes from telephones or PCs by entering account numbers and personal passwords. Telephone access allows information entry and all system commands to be performed via the phone's touch-tone keypad. In addition, more advanced systems deploy speech recognition technology to manage messaging via simple voice commands over the phone.
Integrated voice mail systems typically have a message-waiting indicator (MWI), such as a light on a telephone or icon on an alphanumeric display. A ringing telephone can default to a voice mailbox that delivers an invitation to leave a message; the system then automatically records the message in memory.
The telephone user interface (TUI) of a voice mail system provides the subscriber with voice menu prompting for message management functions, including retrieval and playback of messages, message disposition (deleting, saving, replying to or forwarding messages to others), sending new messages to one or more subscribers and changing the setup of mailbox facilitiesgreetings, passwords, distribution lists and access to live assistance. In addition, the growth of e-mail usage has increased the popularity of the PC screen as a voice mail user interface. Point-and-click techniques, together with labeled icons, make desktop voice message management easier and more efficient.
The popularity of Internet-based e-mail has positioned it as a target for integration/consolidation with voice mail systems and services. These services are particularly well suited for small businesses and SOHO environments that do not have an internal enterprise e-mail server. In addition, the Internet offers a cost-effective means of voice mail networking between diverse voice mail systems using the Voice Profile for Internet Messaging (VPIM) industry standard. Popular Web browsers are also being employed by users for PC access to voice mail servers over the Internet and can be used by messaging system administrators to remotely manage user activities via the Web.
Unified messaging is an extension of voice mail that enables subscribers to access and manage messages such as voice mail, fax mail and e-mail from a single user interface. The goal is to simplify and speed message handling by improving how subscribers receive, reply to and manage messages, regardless of communications medium.
Implementation can consist of a single unified messaging server or multiple servers behind a single user interface. The user interface itself is typically a desktop PC driven by an application software module or integrated software load. The PC-based interface also enables data files to be easily attached and retrieved with any form of message medium, not just e-mail. Several of today's sophisticated unified messaging packages allow subscribers to embed voice messages in fax and e-mail files, view faxes onscreen, be notified of e-mail (by PC or telephone) and redirect e-mail to a fax machine via telephone commands. The telephone is another user interface (TUI) alternative for unified messaging. In addition to listening to voice messages, a text-to-speech option is generally offered for the telephone retrieval of text messages from e-mail or fax.
Terminal Adapters (TAs)
A terminal adapter, also known as an ISDN modem, has two main tasks. One is to adapt the format of the data or voice signal at the R interface (the interface between a non-ISDN terminal and the TA) to the 64-Kbps B channel. The other task is to provide a means of setting up and clearing ISDN calls. Like modems, terminal adapters can be packaged in a variety of ways. Individual basic-rate TAs can be obtained in stand-alone boxes, each with an ISDN port and one or two terminal ports. For central site use, a number of TAs can be mounted in a rack. Finally, a TA on a card can be plugged into a personal computer or workstation.
Communications processors are multifunctional, program-controlled computerstypically called "servers" todaydedicated to communications and serving as control points, or nodes, in networks. In general, a processor performs one or more of three major functions:
A front-end processor serves as a locally attached peripheral device to one or more larger computers, relieving them of the overhead involved in message handling and network control that is required in a communications environment. An intelligent switch routes messages among the network's various end points and participates in the network's control and management, either under the control of a master (usually front end) processor or as a peer of other intelligent switches.
A concentrator controls a community of terminals, clusters of terminals or distributed applications processors. It gathers, queues and multiplexes their transmissions onto one or more high-speed network trunks and participates in the network's control and management, again either under the direction of a master processor or as a peer of other concentrators and switches. Most high-end communications processors perform all three of these tasks.
Virtually all modern telephone networks use digital transmission to connect digital switching offices. At the edges, however, networks still use analog facilities to connect to customer equipment, especially residence and office equipment, such as telephones, fax machines and computers. For the network to carry data (e-mail, documents, spreadsheets or other data types) and digitized fax images, the information must be converted into a continuous (analog) line signal variable along one or more of the parameters of amplitude, frequency and phase.
A device called a modem, a contraction of the terms modulate/demodulate, performs the conversion from data format (1s and 0s) to analog format. Modems are always used in matching pairs and generally conform to ITU-T standards. The unit at the sending end converts information coming from a host, PC or terminal. At the receiving end, another modem converts the analog signals back into data format before acceptance by the receiving devices. In the case of a fax, the receiving modem's output is digitally processed by a DSP in the receiving fax machine to create a nearly identical rendering (that is, facsimile) of the original image.
Modern modems can transmit and receive simultaneously (full duplex) on the public network at speeds up to 56 Kbps. V.34 is a popular international standard for dial-up modems, supporting speeds of 28.8 Kbps. The more recent V.90 standard supports downstream operation at 53.3 Kbps and upstream at up to 33.6 Kbps. The even more recent V.92 standard improves on performance even further, increasing upstream speeds to a maximum of 44 Kbps, under optimum conditions.
Multiplexers combine streams of data from many individual low-speed channels and transmit a combined stream over one high-speed communications link. Multiplexers maximize the efficient use of communications links in a network because users can lease one high-speed line for much less than it would cost to lease many low-speed lines. Multiplexers generally fall into one of two very broad categories:
FDMs are the earliest and least sophisticated form of multiplexing equipment. FDMs divide the allocated bandwidth of a conditioned analog line into independent, permanently assigned lower-speed subchannels that operate on a particular frequency within the spectrum.
TDMs are digital devices that accept multiple digital inputs and convert them to one composite digital output. Rather than divide a communications link into separate channels as FDMs do, a TDM divides bandwidth into time slots while maintaining the integrity of the single channel. Within the TDM classification, there are several different types of devices:
In a simple fixed-format multiplexer, the relationship between the input and the output is fixed. In a more sophisticated multiplexer, it may be varied.
Statistical time division multiplexers (STDMs) contain an added microprocessor that provides intelligent data flow control and enhanced functionality, such as error control and sophisticated user diagnostics. The major difference between TDMs and STDMs is that STDMs dynamically allocate time slots on the link to inputting devices on an as-needed basis, rather than in a fixed-dedicated basis. Therefore, rather than wasting bandwidth when the inputting device is idle, the bandwidth is utilized to serve active devices (that is, devices with data ready to go). STDMs work best when data flow is intermittent; if data from multiple devices occurs simultaneously, one or more devices will have to wait. Unlike TDMs, STDMs have buffers for holding data from attached devices, which enable them to handle a combined input speed (aggregate speed) that exceeds the speed of the communications link.
Drop-and-insert (D/I) multiplexers are commonly used in private networks and in the dedicated facilities portion of public networks. D/I muxes are used to remove one or more of the channels from a multiplexed transmission system or to add more channels to vacant slots in a multiplexed system. On a small scale, they perform a function similar to DCSs, described previously. Add/Drop Multiplexers (ADMs) used in ATM networks are one example.
As mentioned earlier, not all data communications devices speak the same language. Network designers can achieve flexibility and economic rewards by using products from more than one vendor, and equipment manufacturers have responded by developing products that overcome language incompatibilities. Protocol converters and emulators can be hardware-based, software-based or a combination of both and can range from a microprocessor-based circuit board to a front-end processor with the capability to handle conversion functions. Available conversion devices might handle one or many types of conversions. For example, some devices handle only code or physical interface conversions, while others handle protocol conversion, device emulation, and code and interface conversions.
A protocol converter actually changes one protocol to another by stripping down the data from one device and reformatting it according to the rules of a new set of specifications. During the conversion sequence, the protocol converter accepts blocks of data in one protocol, adds or deletes the necessary control characters, reformats the block and calculates the required check characters so that the receiving device gets characters formatted according to its requirements.
For example, in an asynchronous-to-SDLC conversion, the converter accepts a string of characters, eliminates start and stop bits, changes the coding scheme from ACSII to EBCDIC, assembles the characters into a block, and adds appropriate headers and trailers to create complete frames.
The major end-user device in the network is the PC, and it has gained nearly universal acceptance as a tool to perform local data analysis. It is only natural that a user would want to communicate with a network host via such an intuitive device.
PCs and data terminals differ in one major respect: the PC is an intelligent device, capable of manipulating and analyzing data and handling a variety of applications, but the data terminal is not. It was, therefore, often referred to as a "dumb" terminal because its basic function was to serve merely as an interface between a human operator and a host computer. Even "intelligent" terminalsthose that can perform some operations on collected datado not offer the sophisticated capabilities of a PC.
Management and Control [return to Table of Contents]
As the voice and data worlds merge, the line between telemanagement and network management software systems blur. Computer manufacturers are incorporating telemanagement functionality into network management systems. Telemanagement software vendors, on the other hand, are incorporating network management capabilities, which are generally voice-oriented and designed to provide PBX management interfaces. The PBX market is the focal point for the convergence of the computing and telecommunications approaches to network management. Most major PBX vendors now have a data networking strategy as well as a voice strategy.
Telemanagement Systems and Software
The growing importance of telecommunications networks within companies means that management tools are essential. Telecommunications management networks are increasingly recognized as strategic assets that can increase customer satisfaction, build customer loyalty and develop new business. Telecommunications management systems not only monitor staff activity and customer response, but also help the telecommunications manager optimize the use of the network and identify problems as they arise.
Telecommunications management systems originated in service bureau-based call accounting software, which provided detailed station reports to large firms as a supplement to telephone company bills. A natural evolution of service bureau software was the development of licensed software systems for a client's computer system. Changes in computer and software technology added significant options to computing alternatives during the mid-1980s, when PCs became increasingly popular. During the late 1980s, LANs became even more popular. Today, software developers offer telemanagement solutions for many of these platforms.
Generally speaking, telemanagement system applications are organized into several categories:
Asset management facilitates the management of physical assets, such as equipment inventories and cable and wire resources. Asset management also may be used to manage logical assets, such as software. Process management automates a number of processes (for example, traffic analysis, network design and optimization, directory management, and work order and trouble ticket management), which enable the effective management of a telecommunications network.
Communications Software [return to Table of Contents]
An efficient method of controlling data communications networks utilizes combinations of hardware and software for control purposes. Repetitive tasks that rarely change are best implemented as hardware modules, while dynamic tasks (such as the maintenance of terminal specifications) are best implemented through easily changed configuration software that can be altered without disrupting network operations.
Data communications softwareoften transparent to the user, particularly in large-scale data networkscan be implemented in several layers, requiring a support staff of specialized programmers for its maintenance and design. Communications software resides in PCs, servers, terminal controllers, front-end systems and mainframes. It is almost always required for establishing some phase of long-distance computer operations.
Communications software is used for the allocation, control and management of the following:
Communications software focuses on a number of objectives: communications between various devices, communications between end users (or between end users and applications), communications between databases, communications between applications, and management and control of communications activities. While hardware generally supports line, device and presentation functions, software supports the following:
Recent development efforts have advanced communications software toward the interaction of end-user programs on a peer-to-peer basis. This bilateral interaction makes the diverse connection schemes and host servers typically found in large corporations completely transparent to end users. It also demonstrates the shift away from rigid, hierarchical network control toward distributed control.
Technology Analysis [return to Table of Contents]
Business Use [return to Table of Contents]
In a complex multivendor environment, end users must assume more responsibility for the network's ongoing functionality. They must seek out appropriate solutions for network redundancy, for troubleshooting and verification, and for network management. Choosing the appropriate equipment for network testing, monitoring and control allows users to carry out that responsibility. As the complexity of the network increases, however, individual test devices can prove inadequate. For such environments, the network management system is essential.
In the past, network management was not really management, it was crisis intervention. Nowadays, network managers are more responsible for the control and monitoring of their own systems. Network management systems have developed from many sources: the desire of interconnect vendors for a value-added selling point, the proliferation of easy-to-use management software and the users' needs to get their communications under their own control.
[return to List of Figures]
Network Management Functions
From a technology standpoint, network management can be depicted as the intersection of seven different functional areas.
A minimal network management system consists of:
All network management systems include mechanisms for monitoring network components. When the network management system vendor also manufactures modems, the vendor usually designs the monitoring device as a built-in modem feature, eliminating the need for the user to acquire separate monitoring devices. On other systems, stand-alone monitoring devices must be attached to modems or multiplexers at each remote site.
In most network management systems, monitoring devices examine only the status of the modem or multiplexer, its interface with the equipment, its interface with the transmission facility and the condition of the transmission facility. Information on the modem or multiplexer and its interfaces comes from the presence or absence of signals on various interface leads. Information on the transmission facility comes from the measurement of various parameters, such as signal level, noise, distortion, phase jitter and line hits. If a given interface signal or analog characteristic falls out of specification, the system's monitors will set off an alarm to notify the operator of a failure.
Some network management systems can switch automatically from a failing component to a "hot" standby unit either on receipt of an alarm or on command from the operator. Some systems can also bypass a failed communications line by a call placed automatically over the switched-voice network. Such automatic dial backup procedures require two switched-network calls for full-duplex operation.
As networks become increasingly sophisticated, network management systems grow in complexity. One of the more interesting recent developments is the incorporation of expert systems techniques into network management. An expert systemsoftware that contains rules for making logical inferencescan add a degree of intelligence to a network management system. Such an intelligent system can do more than isolate faults; it can also suggest to an operator the possible causes of those faults, test hypotheses about them and propose courses of action to remedy them.
Multiple network management systems can coexist within a corporate network. Communications networks today may be built from several different types of dissimilar equipment and may incorporate several smaller networks of different types. Full network management may require interconnecting management information from all of these networks into a usable form. One of the frontiers of network management is the construction of systems that can bring together this dissimilar information and integrate it through decision support or executive support systems.
Technology Leaders [return to Table of Contents]
Listed below are some of the leading vendors in both the data and voice communications industries for equipment and services:
Insight [return to Table of Contents]
Movement continues today toward the integration of enterprise voice and data communications networks, spurred-on largely by the interest in voice over data networks, with the emphasis clearly on VoIP. Justifications for converging onto a single infrastructure have progressed beyond toll bypass applications to projections of lower total cost of ownership (TCO), more efficient system management and administration, and enhanced applications use. But QOS issues and lack of consensus on standards continue to be inhibitors, along with a perceived shortage of personnel with expertise in both areas. In addition, with shaky economic conditions still existing globally, enterprises are more carefully scrutinizing any modifications to current infrastructures.
Ray Horak, president of The Context Corp., Mt. Vernon, Washington, developed this report exclusively for Gartner. Context is an independent consultancy that works closely with manufacturers, developers, distributors, carriers and end users across a wide range of technologies and applicationsat both the strategic and tactical levels.
Mr. Horak is an internationally recognized author, technical writer, seminar leader and lecturer. He has written the best-selling Communications Systems and Networks, published by John Wiley and Sons and currently in its third edition, and is senior contributing editor for Newton's Telecom Dictionary. He is a member of the faculties of Network World Technical Seminars and Terrapinn and is a regular speaker at leading industry conferences. Mr. Horak serves on the editorial boards of a number of periodical industry publications and on the advisory boards of several colleges and universities.
The report draws heavily on Communications Systems and Networks, authored by Ray Horak and published by John Wiley and Sons.