What is the definition of data communication


Data has been sent over telephone lines using modulated signals for a few decades. But it was only about a dozen years ago that 'modems' slowly became popular with home computer users. Only a little later, so-called mailbox systems saw the light of day - computers that offered their users information and programs using suitable software and later also made it possible for users to communicate with one another. On the user side, only a modem and a terminal program were required. At that time, there were also many technical and bureaucratic hurdles to be overcome. I gained my first experience as a mailbox sysop when I set up the mailbox for the readers of the computer magazine 'mc' in 1983. It consisted of a 300-baud modem from the Post Office and a QX-10 from Epson (an 8-bit computer with the CP / M operating system and two floppy disk drives) that ran with self-written software. Since then, remote data transmission has stuck with me, be it as a user of larger mailbox systems (locally in Munich CUBENet or worldwide Compuserve) or while working at the Munich University of Applied Sciences. For about two years now, there has also been a change in mailboxes. They offer access to the global Internet network via their dial-up computers. Modem technology has not stopped either, data transfer rates are now 56,000 bit / s and more. And the problems of the newcomers have basically remained the same. This book is intended to help you get started with technology. But even the professional may still find something new. Data communication via ISDN has been left out on purpose; this complex technology will be dealt with comprehensively in another book from this publisher.

1. Data communication

In data communication, computer data is transmitted over direct cable connections, telephone lines, or radio. There are two communication terminals, mostly computers, at both ends of the line. This chapter will first clarify some basic terms and then deal with the various options for transferring data via telephone and ISDN connections.

1.1 Introduction

Data communication takes place in spatially limited as well as over long distances. There are different levels of communication:
  • direct connection of two devices (computer-printer, computer-computer in the same room, etc.)
  • Communication of several devices within a limited area (within a building or building complex) -> LAN (local area network), local networks
  • Communication via public services (post, open networks) -> telecommunications, WAN (wide area network)

Examples of data communication applications:

  • Information transport to the place of need (direct, suitable for processing), e.g. B. Connection of locally computer-controlled processes to central control and evaluation computers
  • Optimal distribution of tasks between local workstations and the central mainframe computer (only tasks that exceed the capacity of the workstation computer are performed on the mainframe computer)
  • Access to centrally stored data from many workstations (LAN, the central computer is called "server")
  • Shared use of special peripheral devices from the workstations (e.g. printer, plotter, etc.)
  • Optimal use of the computer capacity through random access to a currently free computer
  • Information exchange (data / programs) between different workstations of a LAN ("client server")
  • Access to problem solutions available elsewhere, e.g. B. other data centers, databases
  • Access to information services, e.g. B. screen text, mailboxes, WAN networks

Basic concepts: message and information

Ask:What is "information"?
 What are "messages"?
 How is information passed on?
 How is information presented?

In addition to energy (and matter), information is a second basic quantity of universal importance. Their independence was only recognized late, as their transmission is always linked to energetic or material carriers. Information is obtained through news. The definition of information in computer science does not entirely coincide with colloquial usage.

    Information is the knowledge of something.
    Messages are used to present information.

In other words:

    The abstract information is communicated by the specific message.

Messages and information are not identical; in particular, the same message (with the same information) can have different effects on different recipients. But there are also messages that subjectively contain no information.

    Example: For which message is the information larger?
    a) On July 1st the temperature was higher than 25 degrees.
    b) On July 1st the temperature was 29 degrees.

With a) there are only two options (smaller / greater than 25 degrees), with b) there are theoretically any number of options. So with b) the information is bigger. It follows that information has to do with the number of possibilities.

    Example: How do I get to my girlfriend?
    Easy way: She lives on the same street. There is only one choice (go right or left).
    Complicated way: there are several branches; At each fork you have to decide whether to go right or left.

The information content of a message can therefore be determined and is determined by the number of (right-left) decisions and is measured in "bits". 1 bit corresponds to a decision:

Easy way:Information 1 bit
Complicated way:2 forks in the road -> 4 possibilities, 2 bit
3 forks in the road -> 8 possibilities, 3 bit
4 forks in the road -> 16 possibilities, 4 bit
1 bit is the amount of data that is transferred with the answer to a yes-no question.
1 bit is only transmitted with a binary character if the two characters in the character set are equally likely. In all other cases less than 1 bit is transmitted.

The Shannon information measureH, or the amount of data for short, is defined as:

The following applies:
N = Total number of characters used
pi = Probability of occurrence of the character i
ld = logarithm to base 2

If all symbols are equally likely, the formula is simplified to: H = ld (N) bit

ld = logarithm dualis: y = 2n -> n = ld (y)
Conversion: ld z = lg (z) / lg (2) = ln (z) / ln (2) (any base)

Is the number of characters N = 2n, so be n bit transmitted. You could call this a n conceive of successive answers to one yes-no question each.

Each end of a branch (leaf) then corresponds to a character, so there are 23 = 8 characters that can be transferred with this.

Presentation of information

After the abstract information now to the concrete representation of the information, the message. Here are a few definitions:

Compilation of symbols (signs) for the transmission of information.

Element of a symbol or character set. This supply is a fixed, finite set of different symbols (= elements of the set). The difference between symbol and sign is quite subtle. A symbol is a sign with a specific meaning.

An orderly store of symbols.

A series of "related" characters that are viewed as a unit in a certain context.

Alphabet: A, B, C, D, E, F, ..., X, Y, Z

Messages can be represented physically by signals, e.g. B. sound waves, electromagnetic waves, currents, voltages, light, etc. A distinction is made between digital and analog signals.

Analog (continuous value) signals:

  • Continuous assignment of a physical variable to the information content of a message, e.g. B. Temperature <--> height of the mercury column, pointer of a measuring instrument
  • mostly very clear
  • any intermediate values ​​can be displayed
  • neighboring values ​​are often difficult to distinguish

Digital (discrete-value) signals:
Here there is only a finite number of possible states of a physical quantity (in the simplest case two), e.g. B. Numeric display of a measuring instrument, sequence of digits. Since most physical quantities are analog, quantization is often carried out, e.g. B. a voltage between 6.5 V and 7.5 V quantized to 7 V.

A distinction is therefore made between analog technology and digital technology. In the digital computers normally used, digital technology with only two possible values ​​is used. These two signals (usually represented by "0" and "1") must be recognizable from the signal in some way. Depending on the physical size used, there are often several clear options for the "0" or "1".

1.2 Basic concepts of data communication

1.2.1 Basics

The core of data communication is the transport of data, the data transmission from a sender to a receiver via a transmission channel. A distinction is made according to the transmission direction possible on a connection:
  • - Simplex operation: unidirectional channel
  • - Half-duplex operation: bidirectional alternating
  • - Full duplex operation: bidirectional at the same time

The data to be transmitted are sent to the transmission medium in the rhythm of a send clock. In order for the information to be correctly retrieved, the received signals must be sampled at the correct point in time. The receive clock must be synchronous with the send clock. Typically the data is converted into a serial bit stream; H. a byte is output bit for bit at a specified data rate (= time interval between two successive bits) -> (bit) serial interface. Different transmission speeds are used in practice. The following values ​​are specified:

50 75 110 150 300 600
1200 2400 4800 9600 14400 19200
28800 33600 .... Bit / s (BPS)   

1.2.2 Serial transmission

Let's first take a look at how serial data transmission works, although the telephone line shouldn't play a role at first. As you may know, a computer stores data in the form of bits, which are the smallest units of information. Such a bit can only assume two states, which can be equated with "yes / no", "0/1", "current / no current". Typically, however, groups of bits are combined into a "word" and processed in parallel by the computer, which makes processing faster. Typical word widths are 8, 16, 32 or 64 bits. 8-bit words are also known as "bytes". Letters, digits and special characters are encoded for processing; H. a numerical value is assigned to each print mark. This assignment is standardized so that the "A" also appears as an "A" on all computers. A code that was originally used for teleprinters has become established for data transmission: ASCII (= American Standard Code for Information Interchange; in German: American Standard Code for Information Interchange).

This code occupies seven bits, and the characters are usually sent in one byte, whereby the eighth bit is often used for data security, i.e. H. to detect transmission errors. The eighth bit was used to expand the character set in the IBM PC-compatible computers that are widely used today.

Since the computer processes the data in parallel, it first needs a so-called "serial interface" for output, which outputs one byte bit by bit serially. For example, the letter "A", which is stored in the computer in the form of the associated ASCII code as the numerical value 65, is transmitted as a sequence of the eight bits 01000001. Each character is preceded by a start bit, which always has the value 0. Since the line is always 1 in the idle state, the receiving module can recognize when a character is arriving. A check bit (parity) can then follow the data bits. At the end there are 1 or 2 stop bits, which are always 1 and thus form a separation from the next start bit.

A character string then consists of a sequence of data bits which are framed by start and stop bits for each character. There can be pauses of any length between two consecutive characters, since the beginning of a character is clearly recognized by the start bit. This is why this form of transmission is called "asynchronous".

The asynchronous transmission reduces the transmission rate, since for z. B. 8 bits of information 10 bits are sent over the line. Another possibility is the transmission of data blocks of several hundred bytes without a pause between the individual characters. A few filler bytes then have to be sent at the beginning of the block so that the receiver can synchronize with the data stream, but then the data is transmitted without redundancy. Such a transmission is called "synchronous".

In order for the receiving block to be able to recognize the beginning of the individual bytes, synchronization must take place at the beginning of the data transmission. This is done by transmitting some synchronization characters (e.g. ASCII-SYN), after which the receiver locks into place. If there is no data waiting to be transferred, the hardware automatically generates SYN characters so that the synchronization does not break off. The synchronous transmission takes place in blocks. The data block is usually terminated by a block protection character (checksum, CRC) and an end-of-block identifier.

Data block Block-

1.2.3 Transmission methods

When transmitting the individual bits, a distinction can be made between two methods, the application of which depends on the transmission medium. If you have a cable connection available, you only have to represent the digital levels in terms of voltage or current levels. This is then called "baseband transmission". The transmission line assumes two (or three) states (levels) depending on the binary values ​​to be transmitted. There are various codings to map the binary values ​​to the line states, which are selected according to various criteria. There are codes from which the transmission clock can be recovered, e.g. B. the "Manchester coding", in which a signal edge (0-1 or 1-0 transition) is generated at the beginning of each bit. In the case of a logical "1", an additional signal edge is then generated in the middle of the bit. In addition, with this coding, due to the constant change, the signal has almost the character of an alternating voltage and can thus be sent over long distances by conventional amplifiers for analog signals. The signal is of course "smoothed" a bit and has to be regenerated at the destination. With ISDN connections, the old lines of the analog telephone connections are reused from the customer connection to the exchange. Here, too, it is important that the signals get through the cables and amplifiers used. The HDB3 code (High Density Binary 3 Code) is used here. It is a more pseudo-ternary code in which the line has three states: L, O, H. The representation of a logical "1" is always alternating with L or H. The logic "0" is a bit more complicated: one to three successive zero bits are represented by a 0 level, but the fourth zero bit is then L or H, precisely the opposite of the L or H level sent last (on H follows L and vice versa). On average, this coding also has no direct current component and can thus be processed by conventional amplifiers.

Baseband transmission cannot be used when transmitting via (analog) telephone lines or by radio. In this case, the binary values ​​are modulated onto a higher-frequency signal (sine carrier). A modem (MondulatorThemodulator) necessary. In the simplest case, a modem converts the serial bit stream into tones of different pitch, for the "0" a lower tone and for the "1" a higher tone (modem = DÜE = data transmission device; the computer is called DEE = data terminal equipment).

Public telephone network_____

In the first generation devices, the telephone receiver was placed on an "acoustic coupler" with a loudspeaker and microphone. This "whistled" the data into the telephone line in this way. At the other end of the line, the audio signals were then demodulated and converted into digital information. Of course, you can save yourself the acoustic detour and feed the signals directly into the telephone line, which is the norm today. The following figure shows the schematic structure of a modem.

There are modems for a wide variety of transmission rates; you can learn more about this in the following section. Modems with 28800 BPS (= bits per second) are currently standard. These modems can also automatically downshift to lower rates. Some modems also comply with the American Bell standards 103 (300 BPS) and 212A (1200 BPS). In addition to a computer and modem, you also need a data transfer program.For many types of computers there are also modem modules that are installed in the computer.

A modem primarily provides:

  • Conversion of the digital signal coming from the data terminal equipment (DTE) into a modulated signal (-> modulator)
  • Recovery of the digital signal from the modulated signal (-> demodulator) In addition, a modem also performs the tasks of interface control and network control. Among others, these are:
  • Electrical termination of the telephone line (or the transmitter interface) so that there is no difference between the voice and data connection from the network (level, frequency range, impedance, etc.)
  • When using the telephone, switch between telephone and modem and vice versa
  • For telephone dial-up connections, dial the partner number and initiate the establishment of the connection
  • Establishing and clearing the data connection (modulation and compression methods, "training" of echo compensation)
  • Monitoring of data transmission on the analog side (signal level = carrier, line DCD)
  • Control of the modem depending on interface signals and provision of status signals of the transmission (CTS, RTS, DTR, DSR, ...)

1.2.4 Transmission parameters

With asynchronous serial transmission, the data bits are framed by start and stop bits. The number of data bits / word, the number of stop bits and any parity bit to be generated can be set. So there are the following parameters:
  • Number of data bits (5 .. 8)
  • Parity bit (none, even, odd)
  • Number of stop bits (1, 2)
  • Transfer rate

The standard setting is 8 data bits, no parity, one stop bit. With the data rate one could assume that the modem (data transmission device, DCE) and computer / terminal (data terminal device, DTE) move at the speed that the modem can handle on the transmission path. Later we will refer to data compression and data backup procedures, which can increase the effective data throughput. With certain modulation methods, a fallback to lower rates is possible if the connection is poor. In modern modems, the data rate between DEE and DCE is therefore set to a certain value. There are several options:

  • One-time defined value (hardware wiring, software configuration)
  • Automatic setting to the transmission rate of the analog connection (modem defines the data rate)
  • Automatic setting to the transmission rate of the DEE-DÜE connection (computer determines the data rate)

Usually the first or the last option is used. Many modems automatically recognize the data rate from the control command (character string "AT"). Due to the data compression, the effective data rate can also be higher than the analog data rate, which is why the DEE-DÜE rate must then be selected higher (e.g. 19200 BPS between DEE and DÜE with V.32 (9600 BPS)). Modem and computer communicate about readiness to send / receive either by software by alternately sending a stop and start character (XON / XOFF or ACK / NAK) or by hardware via the CTS / RTS lines. The connection to the telephone line is normally controlled by the DTR line.

1.2.5 Baud and BPS

Discrete-time signals are characterized by a minimum time interval between two successive changes in the signal value. At isochronous For digital signals, the points in time of the change from one value to another lie in a fixed time grid with the distance T.

The walking speed is 1 /T. Your unit is baud=1/s. (named after the French E. Baudot), which comes from telex technology. In baud the number of information changes per second is given. Here is an example: We define two binary states 0 and 1 for our transmission path (in this case it should be a simple cable). The 0 should correspond to a voltage of 0 volts, the 1 to a voltage of 5 V. Here the baud rate is the same the number of bits transmitted / second.

Depending on the number of signal values, a distinction can be made between:

  • Two-valued digital signals.
    Digital signal in which the signal elements are binary (as above).
  • Multi-valued digital signals.
    The signals can take on more than two values, e.g. DIBIT - two bits per signal value.
With modern transmission methods there are more than two signal values ​​and the transmission speed is therefore specified in bits per second (BPS). The data rate (BPS) can, however, differ from the step speed (baud).

Because we have an analog transmission line, we can also make another arrangement: Four different voltage values ​​are used, 0 V, 5 V, 10 V and 15 V. The bits are now combined into pairs (dibits). The assignment is z. B. chosen as follows:

00 ---> 0 V 01 ---> 5 V 10 ---> 10 V 11 ---> 15 V

Now twice as many information bits can be transmitted with the same baud rate (!). You then have z. B. 300 baud, but 600 bps. The method can be extended by combining 3 or 4 bits into one unit. Other methods of transmission technology will be dealt with later.

The question now arises as to how high the data rate can be increased for the telephone line. The telephone connection has a permissible frequency range of 300 Hz to 3400 Hz. Due to the attenuation, a maximum of only around 2500 Hz can be used. The maximum baud rate is twice the limit frequency, i.e. 5000 baud.

In the case of an analog connection, noise is also included. The telephone line has a dynamic range of -40 dB to approximately -3 dB in order to avoid crosstalk between individual lines. Is vs= 1 / T the step speed (signal rate) in baud, then applies to the transmission rate (bit rate)

vb= ld (n) * vs,

in which n indicates the number of signal values.

Each transmission medium or system only transports a finite frequency band. The width of this frequency range is called bandwidth.

For this reason, the transmission rates of channels are also limited. Is H the bandwidth of the channel in Hz, V. the number of signal values ​​for coding applies to a noise-free channel (Nyquist, 1924):

R = 2 * H * ld (V) (in bits / s)

For a channel with random noise the following applies (Shannon, 1948):

R = H * ld (1 + S / N) (in bits / s)

in which S / N the signal to noise ratio is (usually 10 * log (S / N)). For a good connection you can set around 35 dB here. This results in a maximum capacity of approx. 30000 BPS.

What is a message cube and what does it say?

The message cuboid describes that within a transmission time Tü a maximum amount of general information can be transmitted. It is H0 the decision content of an instantaneous value and B. the bandwidth of the transmission channel.
A continuous source does not send out individual, discrete characters, but delivers a stochastic (random) function s (t). Their distribution density is continuous. The instantaneous values ​​are infinitely finely graduated in contrast to the finite number of steps of a discrete source. In physics and technology, however, there is no such thing as an infinitely fine resolution. There is always interference, for example in the form of noise.

In the case of signal superposition (power Ps) and noise (power PN0) counts as decision content H0 of a single instantaneous value in bits (formula 1)

The prerequisite for this is that the signal and noise are normally distributed, i.e. that their frequency distribution corresponds to a Gaussian bell curve. The signal-to-noise ratio is the decisive factor. The square root of the quotient of total power (signal and noise) and noise power can be interpreted as the number of distinguishable amplitude levels when compared with H.0 = ld (1 / P) [bit] is drawn for discrete information sources.

The information content H can maimal the value of the decision content H0 to reach. It depends on the distribution density of the signals. According to C. E. Shannon can capacity C. of a bandwidth channel disturbed by normally distributed noise B. be calculated. Additional noise is added to the signal in the channel. The total noise at the sink is n (t) and its power PN. For the content of the decision, in the above context for PN0 the noise power PN to use. According to Shannon's sampling theorem, the entire content of the signal function s (t) can be recorded with samples at a distance of T = 1 / (2B). This means that the maximum possible number of independent function values ​​can be transferred per unit of time.

The channel capacity for this is given in formula 2 above (unit bit / s). Within the transmission time T0 the maximum total information content HMax can be transferred according to formula 3. This product can be represented as the volume of a cuboid, the edge lengths of which are determined by the three parameters mentioned.

The amount of information to be transmitted HMax can be represented according to K. Küpfmüller as the volume of a cuboid, the so-called message cuboid. A certain amount of information can in principle have any values ​​of

  • Transmission time,
  • Bandwidth and
  • Decision content (depending on the signal / noise ratio)
be transmitted if the volume of the message cuboid remains constant. A message flow H 'can thus be adapted to a transmission channel by reshaping the message cuboid while maintaining its volume.

If the decision content is kept constant, the bandwidth and the transmission time can be changed under the condition that the product of the two remains constant. According to Küpfmüller, this relationship is also known as the "time law of electrical communication". It says that the time it takes to transmit a message must be greater, the smaller the available frequency bandwidth.

If the duration of a message is to remain unchanged, the bandwidth and the decision content can be changed under the condition that the product of the two remains constant. This means that a message can be adapted by frequency band compression to a channel (with C> = H ') which has a bandwidth that is too small for the signal, or by frequency band expansion to a channel which has too low a signal-to-noise ratio.

Echo cancellation

At high data rates and wide area connections, echoes are a problem in data transmission. For telephone connections over 2000 km away, echo barriers are looped into the connection. Each of the two participants in a telephone connection not only hears their voice directly (via the air and as structure-borne sound), but also as an echo from the partner device. With short signal propagation times (short-range connections) the echo cannot be heard at all or only as a diffuse reverberation. With longer signal propagation times (long-distance connection, especially with satellite connections), however, there is a clear echo that irritates the speaker. The echo barriers work voice-controlled and only release the "speaking direction". The echo suppressors are a serious obstacle for data transmission, since only half-duplex transmission is then possible.

The modem can turn off the echo locks by sending a 2100 Hz tone. But now the two modems connected to each other have to cope with the echoes of the transmitted signals:

  • Near echo: The echo that is generated at the point where the modem is analogously coupled to the telephone line.
  • Distant echo: The echo that comes back from the partner station.

With pure frequency modulation, the filtering is easy, since both modems only have to work on different frequency bands (answer / originate mode). With quadrature amplitude modulated signals, both stations use the full bandwidth of the voice channel. The echo compensation must be adapted to the line for each connection (level, transit time). Therefore, such modems usually use adjustable filters or digital signal processors. When establishing a connection, both modems "train" each other by alternately sending short, defined signal sequences. If the line quality is poor, a "fallback" to lower data rates is possible.

Principle of echo cancellation

1.2.6 Modulation method

If a station only transmits in one direction at a time and the remote station remains quiet during this time, you can use the maximum bit rate (half-duplex operation). However, if both stations want to send and receive at the same time (full duplex operation), the data rate drops. In order for the data to come over the line, they must, as already indicated, be modulated onto a carrier signal:
    With amplitude modulation (ASK = amplitude shift keying), the amplitude (signal voltage) of the signal is changed, which has a constant frequency. In the simplest case, this is done by pressing the carrier on and off. The base frequency of the carrier is much higher than the number of blanking processes. It is the simplest method, but interruption and zero bits are indistinguishable from one another.
    With frequency modulation, the frequency (pitch) of a signal is changed at constant amplitude (FSK = Frequency Shift Keying). Two different frequencies are assigned to the values ​​"1" and "0". For duplex operation, different carrier frequencies are used for the outbound route (originate) and return route (answer). An interruption (failure of the carrier) can be seen.
    With phase modulation (PSK = phase shift keying), the signal has a constant frequency. Phase jumps are "built into" the sinusoidal oscillation here. Imagine a sine wave. A phase jump then leads to a certain amplitude, which depends on the phase angle, i. H. the beginning of the oscillation of the sine wave is changed by the corresponding phase angle. High transfer rates can be achieved with PSK, but there are also high demands on the hardware.

Summary of modulation types:

But now to the individual transmission standards as defined by the CCITT (Comit Consultativ International T l graphique et T l phonique), today ITU (International Telecommunications Union). The transmission speeds below 2400 bit / s are hardly used today. The reason for the development is actually the above-mentioned limitation of the bandwidth of an analog telephone connection to a bandwidth of slightly more than 3000 Hz. Through constant improvement of the sending and receiving hardware, attempts are being made to achieve ever higher transmission speeds. In modern modems, digital signal processors are often used, which can simulate analog functions (e.g. filters, demodulators, modulators) by software with the aid of digital-to-analog and analog-to-digital converters. These modems can often be adapted to new processes by means of a software update. Since the signal processor can in principle process any analog signal, such modems often also have the option of voice recording and playback. Together with the appropriate computer program, they can then also be used as an answering machine or voice information system.

V.21 (300 bps)

This is the oldest standard that was used in the first postmodems and acoustic couplers, it only has historical significance. The bits are converted into tones of different frequencies (frequency modulation). For two channels you need four frequencies:
    Channel 01180 Hz1850 Hz
    Channel 1980 Hz1650 Hz

This modulation method is still used in a number of systems when establishing a connection, for example to adjust transmission parameters or the final transmission method.

V.23 (1200/75 BPS)

This standard is used for Datex-J transmission, in which the amount of data in the two directions is very different. The data is transferred from the Datex-J computer to the user with 1200 BPS, from the keyboard of the user to the Datex-J computer with 75 BPS. Almost the entire frequency band is used for transmission with 1200 BPS, the 75 BPS just barely get through. Also used for 1200/1200 half duplex.

V.22 (1200 BPS)

Two bits at a time are combined into a so-called "dibit". A phase angle is then assigned to these dibits: 00 = 0 degrees, 01 = 90 degrees, 10 = 270 degrees, 11 = 180 degrees. So 600 states / second (= 600 baud), but 1200 BPS are transmitted. V.22 also applies to 600 BPS (only two phase angles). Projected graphically into the plane it looks like it is shown in the picture. The transmission takes place here in full duplex, i. H. both stations can send and receive at the same time. Both modems send their information on their own carrier:
  • calling modem (originate): 1200 Hz
  • answering modem (Answer): 2400 Hz

V.22bis (2400 BPS)

Here, too, phase shifting is used, but the amplitude is also modulated (quadrature amplitude modulation). In addition, the phase angles 45 degrees, 135 degrees, 225 degrees and 315 degrees are introduced. With a baud rate of 600, 4 bits can be transmitted with each step. In addition, V.22bis is also compatible with V.22, so that 1200 BPS are also possible. The graphics are more complex:

V.27ter (4800 BPS) and V.29 (9600 BPS)

If you double the step rate in the previous procedure (1200 baud), you can transmit 4800 BPS. A further doubling then leads to 2400 baud and 9600 BPS, because 4 bits are always transmitted per step.
However, these methods can only transmit half-duplex, since the entire bandwidth is used. They are mainly used for fax transmission, as the data stream goes in one direction and the recipient only acknowledges briefly.
Since the telephone line is by no means free of interference (crackling, hissing, etc.), a great deal of technical effort must be made to enable the transmission. Fax machines can switch back from 9600 BPS to 7200 BPS or 4800 BPS even with poor lines (fallback). The V.29 phase diagram is shown in the picture.

There are still some versions of V.29 that have an additional return channel with 300 BPS. Despite the half-duplex operation, the receiver can send messages to the sender on this channel (acknowledgment of receipt, cancellation, etc.).

V.32 (9600 BPS)

With a modem, of course, you want duplex transmission. Here, the transmission signal is filtered out of the composite signal with the help of signal processor modules. In this way, the partner's information can be recognized. The modulation frequency here is 1800 Hz. Both stations require a training phase in which only one partner transmits at a time and can thus adjust its echo blocker. The "trellis method" is also used here, in which four data bits are transmitted encoded in five bits. The fifth bit is calculated from the four data bits and has a similar effect to a check bit when coding seven bits in one byte. This results in a 3 dB better signal / noise ratio compared to pure quadrature modulation. The trellis method subtracts the transmitted data from the composite signal to eliminate echoes and thus extracts the received information. Incidentally, this complex procedure is also used with ISDN.

V.32bis (14 400 BPS)

By increasing the step speed again (2400 baud) and adding further phase angles, this fantastic speed is achieved with the trellis method (128 states, 6 data bits, 1 redundancy bit). However, the electronics required here are also considerable. Usually the modems either contain special circuitry or use a digital signal processor. V.32bis, like V.32, has a fall-back mode of 4800 BPS. Some manufacturers achieve even higher data rates (up to 19200 BPS) by modifying the process. At the higher rates, however, data exchange can only take place between modems from the same manufacturer.

V.32terbo (19200 BPS)

By modifying the V.32bis, the transmission rate can be increased to 16800 BPS and 19200 BPS. For this purpose, the trellis method is used again, but the number of bits is increased to 7 or 8, the amplitude and phase jumps being assigned non-linearly in order to facilitate decoding on the receiver side. In this way, non-linear distortions can be better "filtered out".

V.34 (V.fast)

This standard was only adopted in the summer of 1994. It defines a full duplex transmission of 28800 BPS with quadrature amplitude modulation and channel separation through echo cancellation. A dynamic adaptation process should enable the modem to transfer data optimally. The symbol rates are already at the limit of what is possible; depending on the transmission rate, they are:
  • 2400 baud (2400, 4800, 7200, 9600, 12000, 14400, 16800, 19200, 21600 BPS)
  • 3000 baud (4800, 7200, 9600, 12000, 14400, 16800, 19200, 21600, 26400 BPS)
  • 3200 baud (4800, 7200, 9600, 12000, 14400, 16800, 19200, 21600, 26400, 28800 BPS).
Further features of the V.34 standard in brief:
  • Negotiation handshake according to V.8 accelerates the connection establishment. The modems exchange all important information using V.21 modulation. With other modems on the opposite side, training continues as before.
  • Line probing ensures the suppression of line interference. The receiver analyzes specified test signals (150 to 3750 Hz in 150 Hz steps) and provides the transmitter with parameters for signal coding. The line parameters are also measured periodically during a transmission, which makes it possible to react to changes in the line characteristics.
  • Non-linear encoding ensures optimal decoding options at the receiver. The coding at the transmitter is adapted to the line distortion according to the parameters determined during line probing.
  • Precoding and pre-emphasis are used to compensate for amplitude distortions by pre-distorting the signal at the transmitter. With pre-emphasis, the signal spectrum can be amplified or weakened in certain areas.
  • Shell mapping for even distribution of the signal points in the phase star.
  • Rate Renegotiation allows the data rate to be adapted to the line conditions, even during a connection.
  • Adaptive Power Control selects the optimal signal level. On the one hand, as high as possible to increase the signal-to-noise ratio, on the other hand, low enough so that the transmitter does not disturb itself through reverb effects.


At the beginning of 1977 something happened that communications engineers had previously considered impossible: Analog modems exceeded the limit of 33.6 kBit / s. 'Almost as fast as ISDN' was the motto when the 56k technology was introduced. There are now three of them, but not everyone is happy with them. At the beginning of 1997 Rockwell, together with Motorola and Lucent as well as US Robotics, presented two technologies called K56flex and X2, with which analog modems could obtain data with up to 56 kBit / s. Although both methods are based on the same principle, they differed in important details and were not compatible. In February all those involved met again to define a uniform procedure: V.90.

At first glance, it seems that 56k modems are exceeding the limit set by Shannon's theorem, but on closer inspection, Shannon's proof remains valid. The 56k data communication is based on a different principle than that of conventional analog modems. 56K devices take advantage of the fact that the host at the provider and the exchange to which the user is connected are connected via a digital line. Accordingly, the host transmits the data digitally until then; Only in the exchange are they converted into an analog signal - the exchange becomes, so to speak, the upstream line interface of the 56k transmitter.

However, the connection between the exchange and the user is so short that the data must be transmitted in analog form, but not by modulating the phase and amplitude of a carrier signal, but can be sent as voltage values. This enables the higher speeds, but only in the direction from the host to the modem. Conversely, the data is transported using conventional methods, i.e. at a maximum of 33.6 kBit / s.
V.90 modems negotiate connections in 56K mode regardless of the manufacturer of the V.90 remote station called. If the line quality is sufficient, data can then be loaded from a 56K host with up to 56,000 bit / s, regardless of the manufacturer. The 56K requirements have remained: A 56K modem, also known as a client, can only receive data with up to 56,000 bit / s from so-called 56K hosts. 56K clients only establish V.34 connections with one another with a maximum of 33.6 kBit / s. The 56K technology is therefore particularly suitable for Internet providers, but mailboxes are also used in some places.

However, a data rate of 56 kBit / s requires an ideal connection between the exchange and the telephone socket. In practice, however, this connection is usually subject to interference, so that the maximum possible transmission rate is hardly achieved. This has brought V.90, X2 and K56flex into disrepute. This goes so far that customers have been advised to buy V.34 modems because the faster modems would not deliver a higher data rate. No modem standard can guarantee the maximum connect rate, because it depends on the transmission properties that change from line to line.

Summary of the modulation method

ITU recommendationmaximum step speed (Baud)maximum bit rate (bps)Modulation method
V.21 300 300 2 FSK
V.22 600 1200 4 FSK
V.22bis 600 2400 16 QAM
V.27ter 1600 4800 8 PSK
V.29 2400 9600 16 QAM
V.32 2400 9600 32 QAM
V.32bis 2400 14400 128 QAM
V.34 3229 33600 960 QAM

1.2.7 Multicarrier procedure

The multicarrier technology comes from the military application, where one wanted to encrypt the data. Up to 500 carrier frequencies are built up in the frequency band of the telephone. These many bit-parallel carriers enable dynamic adaptation to the state of the line. Disturbed frequencies are masked out, i.e. not used. The many carriers enable data rates of up to 19200 BPS. The best-known method of this type is PEP (Packetized Ensemble Protocol), which has not yet been laid down in any standard. Therefore, there are currently several variants. In principle, similar procedures are used with ADSL.

1.3 Data compression / data backup

1.3.1 Transmission protocols

A certain procedure, a "protocol", is used to transmit the data. If you read a text, some transmission errors do not interfere. It is different with data or programs; here every byte has to be correct. Therefore, the data is transmitted in blocks. The transmission program calculates a checksum for each block, which is also transmitted. The receiving program recalculates the checksum and requests the data block again if the two checksums do not match. Modem traffic typically uses simplex protocols; H. the data flow is only in one direction. In the opposite direction, only acknowledgment signals or blocks are transmitted. More recently, protocols have also been developed that allow simultaneous data transmission in both directions or even make it possible to allow several processes to communicate via a serial connection. These processes (mostly based on HDLC) then lead to network connections.

One of the oldest protocols was "Kermit" with a maximum block length of 94 bytes plus checksum. With Kermit, as with all other protocols, the data is sent in blocks, with the other station confirming each block positively or negatively. Bad blocks are repeated. Since z. B. If the connection is interrupted, data or confirmation are missing, the transmission is aborted after a definable waiting time (timeout). At Kermit, all communication is handled using complete blocks - the confirmation also consists of a block that contains just one usable character. The structure of a Kermit block looks like this:

SOH ASCII character "Start of Header"
LEN Number of characters in the block (from SEQ up to and including BCC) 32 is added to the length specification, so the value is transposed to the range ASCII "#" (decimal 35 = length 3) to "~" (decimal 126) -> printable Character.
SEQ Block number modulo 64. 32 is added again -> range from "" (decimal 32) to "_" (decimal 95).
TYPE Type of block. There are the following types:
    S. (Send init) Start of sending (parameter transmission)
    F. (File) filename
    D. (Data) data
    Z (End of File) End of file
    B. (End of Transaction) End of transmission
    Y Positive receipt
    N Negative receipt
    E. Fatal error
BCC Block check characters
CR ASCII Carriage Return (decimal 13)

Since the block length is communicated to the recipient, the user data can be transmitted transparently.

After Kermit came "X-Modem", which is faster and also sufficiently reliable. This protocol uses a fixed block length of 128 bytes. Since X-Modem always transmits complete blocks, the files can be extended if necessary, which sometimes causes difficulties. As with Kermit, each block is introduced by the ASCII character SOH, but the block structure is different. Since the length of the data block is also specified here, the user data can be transmitted transparently. A disadvantage compared to Kermit is the lack of the file name. An XModem block has the following format:

SOHSEQCFEData block (124 bytes)BCC
SOH ASCII character "Start of Header"
SEQ 1-byte block counter
CFE Complement of SEQ
BCC 1 byte checksum.
Adding the block counter and complement always results in 0FFh.

With XModem, the opposite station does not confirm with a data block, but only with an ASCII character (positive: ACK, negative: NAK). In the end, no complete block is sent, but only the ASCII character EOT. The following figure shows the protocol sequence.

"Y-Modem" is an extension of the X-Modem protocol. The block size is adapted to the line quality (bad line -> small blocks). The maximum block size is 1 KByte; if the line quality deteriorates, the block size is dynamically reduced. If the quality improves, the blocks become longer again. It is possible to transfer several files in one step.

In the meantime, the Z-Modem has become the standard for mailboxes. It has an improved checksum calculation, variable block lengths and is also much faster (at 2400 baud approx. 230 characters of useful information / second). As with Kermit, the file names are also transferred so that several files can be transferred at once. Z-Modem is also able to resume an interrupted transmission at exactly the same point at which it was interrupted and is therefore used almost everywhere for the automatic data exchange between the mailboxes.

1.3.2 Error correction and data compression in the modem

From time to time there are transmission errors due to interference on the telephone line. Instead of dealing with error detection and correction via the software protocol, the modem hardware can also be made "more intelligent". The "Microcom Networking Protocol" (MNP) from Microcom is an error correction method with which a completely error-free transmission is possible even with a disturbed line - as long as both sides master the protocol, which in practice means nothing other than that both modems can transfer the data transmitted using this protocol. Further levels of the MNP protocol also allow data compression in real time, which increases the overall throughput of the modem. For the European area there are protocols according to V.42 (error correction) and V.42bis (compression). The process is transparent to the computers connected to the modem, so no special measures need to be taken. Since the data rate between the modem and the computer can be higher than that on the analog line due to the compression, a higher data rate must be set between the computer and the modem.

There are several classes of the MNP protocol, of which MNP4 and MNP5 are currently the most important.

The MNP1 class corresponds to the BSC protocol, an asynchronous, byte-oriented half-duplex method. Due to the protocol overhead (including start and stop bits), the throughput drops to 70% of the data rate. The memory and processor requirements in the modem are low.

With MNP2 the process works full duplex. Due to the higher processor performance, the throughput is approx. 84%.

MNP3 works with a different protocol. It is now synchronous, i. H. transmitted without start and stop bits. The data block is protected with a CRC check information. In the event of an error, a data block is repeated. The protocol is based on HDLC. The transmission between the modem and the computer is still asynchronous with start and stop bits.

MNP4 adds two new concepts to Class 3: Adaptive Packet Assembly (APA) and Data Phase Optimization (DPO). APA constantly checks the line quality. With good connections, the size of the data packets is gradually increased, with poor quality it is reduced accordingly. This increases the overall throughput. The disadvantage becomes apparent with sporadic disturbances - larger blocks have to be repeated. DPO is intended to reduce the protocol overhead. The modem removes repetitive status and control information from the data stream. The combination of APA and DPO increases throughput to 120%.

MNP5 can also compress the data so that the transmission time is shorter (1.3 to 2 times as fast). For this purpose, the data stream is analyzed in real time and the data is transmitted in compressed form (Huffman method). Of course, if the data is already in a compressed form, that doesn't help much. In the case of pure text transmission, however, the advantage of data compression becomes significant. In addition, the bit stream is made fail-safe according to MNP4.

MNP6 introduces two more features: Universal Link Negotiation, ULN and Statistical Duplexing, SD. ULN should establish the best possible connection between different modems.For this purpose, the lowest data rate is started when the connection is established and it is switched up as far as possible during the training phase. With half-duplex protocols (e.g. V.29), SD simulates a full-duplex connection between modem and computer. Send data is temporarily stored and transmitted in pauses between reception (ping-pong system).

MNP7 is an improvement on MNP5. The Markov method is used, which tries to achieve an optimal compression of the data based on the observation of the data stream. For this purpose, the tables of the Huffman coding are modified again and again. The throughput can be up to 3 times as fast. MNP7 also supports V.42bis (see below).

MNP8 combines MNP6 with MNP7. However, it will not be further developed as it is only of interest for V.29 modems.

MNP9 adapts the protocol to V.32 modems on the basis of MNP7. The data throughput reaches about 300% in full duplex. The time required for the acknowledgment is reduced by not sending a separate acknowledgment block, but rather "strapping" the acknowledgment to a data packet in the opposite direction. In addition, the repeated sending of incorrect packages is reduced. In the previous procedure, all blocks sent up to that point were repeated after an error acknowledgment (acknowledgment and send blocks do not run synchronously, but within a "window", that is, the acknowledgment of each block is not waited for, but rather sent cheerfully until an error message appears) only the bad blocks.

MNP10 promises to be the future standard. This method should optimally cope with fluctuating line quality. Five improvements are introduced:

"Robust Auto Reliable Mode", which is supposed to filter out disturbances when establishing a connection (previous levels abort the establishment in the event of disturbances).

"Dynamic Speed ​​Shift" continuously adapts the data rate to the quality of the line - the highest possible data rate is always used.

"Agressive Packet Adaptive Assembly" improves throughput by increasing the packet size starting with 8 bytes of user data to a maximum of 256 bytes. So far, the maximum packet size was started and then gradually reduced (MNP4 works with fixed sizes: 32, 64, 128, 192, 256 bytes).

"Dynamic Transmit Level Adjustment" adapts the transmission level to the line quality. Connections should still be possible even with a signal / interference ratio of 14 dB.

If you consider the current maximum modem performance of 28800 BPS, combined with MNP10, transfer rates of up to more than 80 KByte / s can be achieved. V.42 corresponds in its performance to the MNP4 protocol, whereby this standard is even MNP4-compatible. However, V.42 has its own, better protocol - LAPM (Link Access Procedure for Modems). As with MNP4, the incorrectly transmitted data blocks are repeated here as well. V.42bis is the data compression standard of the ITU-T; it delivers a compression rate around 35% higher than MNP5 (Lempel-Ziv-Welch compression). A V.42bis modem can also recognize whether the data is already available in compressed form (in most mailboxes the data is already "packed") and only performs the compression on data that can also be compressed. V.42bis requires the error detection of V.42. The procedure is not MNP5-compatible, but can process the error detection of MNP4.

When using data compression, the transmission rate between the computer and the modem must always be set higher than the data rate between the two modems themselves.

Copyright © Munich University of Applied Sciences, FB 04, Prof. Jürgen Plate