An introduction to how data communications works by networking of computers, terminals and modems.

An Introduction to How Data Communications Works by Networking of Computers, Terminals and Modems In Figure 1, we see a small communications network. In this figure, a terminal communicates through a modem, across a telephone line, through a remote modem and into a computer. The terminal shown interacts with the computer.

The flow of data starts at the keyboard when a key is depressed. The keyboard converts the key depression into a string of binary ones and zeros, shown in the example as an ASCII stream of eight bits. The data stream is transmitted over the communications line to the computer. The computer echoes the character back and it appears on the display.

This type of communications is used on most computer time-sharing systems. What you see on the display is actually echoed from the computer.

The important thing to remember is that the key depression converted the keystroke into a string of eight binary bits. Since any character can be expressed in a binary string, it can be given to the modem and converted into "marks" and "spaces" for communications across a telephone line.

The terminal uses a form of communications known as asynchronous. Since the characters are sent as the keys are depressed by the operator, we don't know the time between characters or when the next character will arrive. In asynchronous transmission, each character will carry a timing signal that tells when it begins and when it ends.

Figure 2 is a drawing of the data stream being put into a modem. Since the input can only be marks or spaces, there are two levels on the graph. When the input voltage to the modem is high, we will refer to this as marking or binary one. When the voltage to the modem is low, we will refer to this as spacing or binary zero.

When the terminal is connected to the modem and no data is being entered, it holds the modem input in the mark state. As long as the input is held marking, it indicates that there is no data coming down the line. When a key is depressed, the terminal will immediately put one bit-time of spacing on the line. The one bit-time of space indicates to the receiving equipment that the next eight bi-times represent a character. The first spacing bit is called the "start bit."

The eight data bits must immediately follow the start bit and must be of equal length. The length of each bit is determined by the speed of transmission. If you compare the marks and spaces in Figure 2 against the bit stream for the character "A," you'll see they are the same.

At the end of the data bits the line must return to the marking state. Even if there is another character to send, there must be at least one bit-time of marking after each character. The mark at the end of the character is called the "stop bit." The stop bit ensures that there will always be at least one mark in every character and that bit timing is correct.

The interactive terminal is a good example of how a piece of computer hardware passes data to a modem. The keystrokes in this case were converted into a serial data-stream and bracketed with timing bits. The entire 10 bits of information were then passed to a modem for transmission. The marks and spaces of the character are then transmitted over the communications circuit.

In this example, the data is shown with a total of 10 bits (one start, eight data and one stop-bit). Transmission systems are made with one, one-and-a-half and two stop-bits. Data also can be five, six, seven or eight bits, dependent on the code used. The important thing to realize is that data can be represented by ones (marks) and zeros (spaces).

Modems Interface Digital Signals with Phone Lines

Modems are devices that take the digital signal from a terminal or computer and interface it to a telephone line. The term modem is a contraction of "modulator-demodulator." The modulator transmits the signal over the telephone line. The demodulator receives the signal and turns it back into digital information.

The bandpass or frequency spectrum of the telephone line is about 300 to 3200 Hertz. Basically, the telephone line is not capable of carrying direct-current voltages. If we were to connect a six-volt battery across a phone line, we wouldn't get six volts out at the opposite end. Since the computer uses voltage levels internally to identify binary ones and zeros, it must have a way of converting its signals into a form that the telephone line can understnd. Since the telephone line is able to pass tones (or sound frequencies), the modem will have to convert the binary signals to tones.

Figure 3 depicts a simple modulator (or transmitter). In this case, the binary ones and zeros are fed into the modulator by the computer. Whenever the transmitter detects a binary one it will put a 1200-Hertz tone on the circuit. Whenever the transmitter detects a binary zero, it will put a 2200-Hertz tone on the circuit. The transmitter then takes the two binary states and generates two tones on the communications link.

When the signal is sent down the telephone line, it's picked up by the demodulator or receiver of the remote modem. The incoming tones are detected and converted back to binary data. Figure 4 shows a simple demodulator or receiver. Whenever the demodulator sees a 1200-Hertz tone on the line, it will output to the computer the voltage to represent a binary one. Whenever the demodulator sees a 2200-Hertz tone, it will send the voltage that represents a binary zero. Thus the demodulator takes tones from the line and converts them back into the voltage levels that the computer is able to recognize.

The signals between the computer and the modem are designed to meet a standard called EIA RS-232. This standard will ensure that although the internal operation and signals of computers will vary, the voltages and signals to modems are always compatible.

Computer Turns the Modem On

The modem modulator will not transmit until the computer turns it on. In the case of a multipoint or half-duplex circuit, it may not be desirable to have multiple modems transmitting at the same time. To eliminate interference, the computer must turn on the modem transmitter. When the computer is ready to send data, it will turn on the transmitter by raising the signal Request to Send. When the transmitter sees the signal it will begin putting a signal on the telephone line and, after a time delay, will send Clear to Send back to the computer. At this point the system has turned on the modem modulator and received confirmation that it is ready. The computer may now begin sending data via the Transmit Data lead on the modem.

The operation of the modulator and the computer are interlocked through the use of the Request-to-Send/Clear-to-Send signals. It's important to understand that the transmitter must be turned on and off before it can send data.

In the demodulator there's a signal called Signal Detect (or sometimes Carrier Detect). When this signal is on, it indicates data tones coming through the telephone circuit. When the computer sees Signal Detect, it knows there is binary data coming in on the lead Received Data. In effect, the Signal Detect qualifies the Received Data as binary signals rather than noise on the telephone line. When the Signal Detect is off, it indicates that the computer should ignore the Received Data.

First Signal Is Ring Indicator 

Assuming the modem is connected to a standard dial telephone line, the first signal on the modem is Ring Indicator Whenever this modem line is called, the ringing signal from the telephone company central office will be detected. If it's a valid ringing signal, the modem will turn on the Ring Indicator so the computer will know someone is calling. In some modems the Ring Indicator will go on and off with the incoming ringing signal (usually two seconds on and four seconds off).

When the computer wishes to have the modem answer the incoming call it will raise Data Terminal Ready. The modem will see Data Terminal Ready and take the telephone line off-hook and connect it to the Transmitter and Receiver. After the line is off-hook and connected, the modem will send the Data Set Ready signal to the computer. As long as Data Terminal Ready is high, the modem will stay connected to the line.

When the data transmission call is completed and the computer wishes to disconnect, it will drop the signal Data Terminal Ready. With Data Terminal Ready off, the modem will cause the telephone line to go on-hook. The modem will also respond by turning off the signal Data Set Ready.

This brief overview was a look at an ideal modem. In most cases the computer may not turn off the control signals. Some terminals don't even use the control signals. The signals required by the interface depend on the computer, the modem and the type of telephone line (leased or switched). It's usually necessary to investigate the manufacturer's manual when troubleshooting the interface.

The RS-232 interface sets the industry standard of the electrical connection between a business machine Data Terminal Equipment (DTE) and a modem Data Communications Equipment (DCE). The standard has been adopted through the Electronic Industries Association (EIA), which sets various interface standards for equipment built in the United States. The EIA standards are not mandatory, but are designed to allow easier integration of components. Manufacturers who wish to interconnect their products should follow the EIA standards wherever possible.

Standard Sets Signal Voltage

RS-232 is the EIA designation for a DTE-DCE interface. The connection for the RS-232-C interface is made via a 25-pin miniature connector. The standard sets the voltage levels for signals across the interface. Signals must be greater than three volts and less than 15 volts. Polarity of the signals determines the function. For control signals, a positive voltage indicates an "on" or turns a function on. A negative voltage indicates an "off" or turns a function off. For data signals, a positive voltage indicates a binary zero or a spacing condition. A negative voltage indicates a binary one or a marking condition.

The RS-232-C interface is the one most-commonly used in data communications, but there are others that can be used:

RS-449--A new EIA standard that is being proposed to replace RS-232. In theis standard the data lines have been reconfigured to allow speeds up to 56 kilobits per second. The lines have been extended to allow distances over 1,000 feet. The RS-449 standard also uses a new 37-pin connector.

V.35--A high-speed interface that makes use of balanced lines for speed up to 64 kilobits per second. The interface is generally used into wideband modems or T-carrier systems.

Current Loop--An interface that uses a pair of wires to pass data between two devices. The voltage on the wires is reversed to show the difference between mark and space. Current loop comes in 60 and 20-milliampere versions. In the connection of current loop, one of the devices must supply the current (the source). The other device works off the supplied current (the sink). Current loop is used where terminals are directly connected to a system without a modem. It generally can achieve speeds of 19,200 bits per second and distances of 2,000 feet.

In. addition, manufacturers develop their own interfaces for specific computer or terminal connections.

The easiest type of modem to understand is the 202 series. The numbering of various modems in the United States usually follows the style numbers of Western Electric products. In this case, 202 series is a group of 1200-bit-per-second modems that can be used on dial or private lines. The 202T indentifies the private-line modem and 202S is the dial-line version.

To understand how the 202 transmits its data, we should understand the frequency response of a basic telephone circuit. The 202 modem will modulate the binary ones and zeros as two tones within the allowable spectrum of the voice line. Whenever the 202 sees a binary one from the computer, it will transmit a 1200-Hertz tone on the line. Whenever the modem sees a binary zero from the system, it will transmit a 2200-Hertz tone on the line. The modem shifts between two frequencies to indicate the shifts between marks and spaces in the data. This type of transmission is referred to as frequency shift keying (FSK). In a frequency-shift-keying selbeme, the area between the two frequencies is unusable. However, the areas outside the two frequencies could be used for additional data channels by using different tones.

When using frequency-shift keying, there are two limitations. First, the transmitted data speed should not exceed the lowest frequency tone used. Second, the higher the speed the wider apart the tones must be. To transmit 1200 bits per second, the 202 has been designed to operate wih the 1200 and 2200-Hertz tones. Because the 202 type of modem uses a major portion of the bandwidth available, it is usually limited in half-duplex operation on a dial line. In this case, the 202 must turn its transmitter off to allow the remote modem to respond. If we want to transmit full-duplex, we have to use a full-duplex private line.

Less Bandwidth Used at 300 b/s

The 103 type of modem is a full-duplex 300-bits-per-second unit. There are many different types of 103 units, but they are usually compatible in operation. The 103 series uses the FSK modulation scheme, which through the use of a slow speed (300 bits per second) is able to use closer-spaced frequency pairs. This allows the modem to use a smaller portion of the available bandwidth.

The 103 uses four frequencies (or tones) to transmit data. The modem at one end of the line will use 1070 Hertz and 1270 Hertz as its mark and space frequencies. The modem at the other end of the circuit will use 2025 Hertz and 2225 Hertz as its mark and space frequencies. By dividing the telephone into a high and a low channel, each modem can transmit at the same time (full duplex).

The decision as to whether the particular modem uses the high or low channel is made by the call origination. When a 103 modem is called, it automatically answers with the high-end frequencies (called F2). The modem that placed the telephone call automatically uses the low-end frequencies (F1). The main use of the 103 type has been in time-sharing systems with interactive terminals. Because these interactive terminals require the character to be echoed back from the system, they need the full-duplex capability of the 103 type of modems.

The Moem Transmits at 2400 b/s

The 201 type of modem is designed to transmit data at 2400 bits per second. Because of the bit rate, the 201 type is unable to use the FSK modulation method. At 2400 bits per second an FSK scheme would require a minimum frequency of 2400 Hertz and a high frequency of almost 5,000 Hertz.

The 201 type of modem uses phase-shifting techniques to overcome the limitations of voice-grade telephone lines. This modem is capable of 2400 bits per second in half-duplex mode on dial telephone circuits. The modem also can be used in the full-duplex mode on a full-duplex circuit.

In the speed ranges from 2400 to 14,000 bits per second, modem manufacturers tend to adopt their own modulation schemes. This can lead to compatibility problems on mixed-vendor networks. The important thing to remember is that modems operated on voice-grade telephone lines at speeds above 2400 bits per second generally use a variation of phase modulation. The actual modulation technique can be determined by either the manufacturer of the equipment or by a standards organization.

The Codex CS 9600 modem uses a modulation scheme called Quadrature Amplitude Modulation (QAM). The modem has 16 phase shifts that encode part of the data and, by adjusting the strength (amplitude) of the signal, the balance of the data is modulated onto the carrier. Through this combination of phase and amplitude modulation the CS 9600 is able to send 9600 bits per second.

Remote Diagnastics Are Included

Another feature manufacturers are incorporating in the higher-speed modems is extensive remote diagnostics. Generally, a narrow-band FSK signal is added to the modem. This extra channel transmits in the 150 to 300-bits-per-second range and is connected to a microprocessor in the modem. At the main computer site a computer polls the remote modems and checks their status. The microprocessor in the modem can report back changes in interface signals or changes in the communications-link signal. The main-site computer can then determine changes in the communications network before a failure occurs. Through the diagnostic system, a technician can test the modem by looping it back or running bit-error tests.

Modem compatibility can be a very serious problem on large networks because

of lack of standards in one case and inflexible standards in others. In the US there are no formal standards. Instead, manufacturers either generate their own or follow the Western Electric modems. If a manufacturer produces a 4800-bits-per-second modem, it may say it's compatible with the Western Electric 208. This lack of formal standards is a problem, because if the vendor is using a non-compatible modulation scheme you may not be able to mix modems on your network. On the positive side, the lack of standards allows a quick evolution to better, lower-error-rate transmission schemes.

In most of the world, manufacturers follow the modulation schemes set down by the Consultative Committee for International Telegraph and Telephone (CCITT). Because networks cross into multiple countries, you must be able to buy compatible modems from each telephone company. CCITT standards allow each country to suuply modems to a common standard. Since most telephone systems are government controlled, foreign operations usually require the installation of a modem that follows a CCITT standard. The use of the CCITT standard makes it easier to find compatibility between modems but also places some restrictions on using state-of-the-art equipment.

The CCITT standards are also not always compatible with the Western Electric modems. For example, referring to the 202 type of modem, it should be the same as the CCITT V.23, but in reality it is not. The V.23 modem uses 1300/2100 Hertz as the carrier while the 202 uses 1200/2200 Hertz. Both of these modems are capable of 1200-bits-per-second transmission speed.

Modem compatibility also becomes an issue as newer and higher-speed modems become available. Currently, manufacturers have available modems that operate at speeds up to 16,600 bits per second, but since there is no CCITT standard to cover this range, the modems can't be used in all countries.

Point-to-Point Doesn't Share

When a terminal is connected to a computer in a point-to-point arrangement, the terminal has its own port that's not being shared with any other devices. The connection to the system may be permanent or via a dial telephone line. If the terminal is used in a data-entry situation where a form is displayed and the operator fills in data, then a line-by-line interaction is not practical. Because the terminal may have to be set up with protected fields and highlights, it would be easier to fill the screen in one transmission. After the operator edits the data, the computer can read the data in one block.

The type of protocol that will handle this function is called point-to-point. In this method of communications, the terminal or computer will send data in blocks rather than unformatted character streams. The use of point-to-point adds a higher level of error checking because it allows the data to be retransmitted if it's incorrect. This is helpful, because the operator is ensured that the data entered is what the system receives.

Same Protocol Can Return Data

The terminal can use the same protocol to transmit data back to the system. This transmission mode allows block-mode sending and checking of data. The point-to-point communications system can also be used for transmission of data between two computer systems.

When terminals are tied together over a multipoint circuit, there must be a protocol established to determine who can transmit at what time. Without any type of protocol, a multipoint line would be chaos. To describe how the multipoint protocol works, we must first examine a multipoint circuit. For example, assume there are four terminals connected to a computer over a single circuit. In this connection, the terminals are called tributaries or slave stations. The computer is referred to as the master.

By designating the computer as the master, it assumes the responsibility of the network manager or traffic cop. It will be the computer that will "poll" each station to see if it has any traffic. In effect, the computer will ask "Do you have any messages?" and the terminal will respond "No." The computer will poll each station in turn. If a terminal does have traffic, it will respond to the poll by sending the traffic to the system.

Added Features Improve Protocol

Once a multipoint protocol is developed, it can be improved through the use of additional features.

Group Polling--If the amount of terminals becomes too high on a circuit, it could take too long to poll the list. The use of group polling allows the system to poll terminal groups with one message. If one or more terminals had traffic, when the poll was received the first terminal would respond with its message, then the second terminal would respond with its message, and so on.

Broadcasting--This is the capability to send one message that would be received by all of the terminals on the line.

Group Select--This is a broadcast to a group of terminals. Only the terminals that are in the selected group will receive the message.

Poll Contention--During periods of low traffic it may take up too much system overhead to continue polling. In this case, a contention protocol can be used. After polling all of the stations and receiving no traffic, the system can invoke the contention mode. At this point the system stops polling the line. If a terminal wants to send a message to the system and it knows the contention mode has been invoked, it will send its address. When the systems sees the contention, it will resume polling, starting with the station that contended. After the network is cleared of traffic the system then will invoke the contention mode again.

Synchronous Transmission is Used

Earlier we noted that every time we send a data character with asynchronous format, we are sending a start and a stop bit. In the interactive terminal, these bits are necessary for timing the characters. In the point-to-point or multipoint protocols, the data in the blocks are contiguous and individual character timing is not necessary.

If we could eliminate the start and stop bits in a stream of characters it could save up to 20 percent on the volume of bits transmitted. The synchronous communications protocol eliminates the use of start and stop bits on each character through the use of a synchronization charac frame.

Synchronous transmission allows the system to increase throughput by eliminating the need for start and stop bits, but the increases in throughput can only be realized if the data being sent is in long blocks.

To sum up the synchronous communications method, it allows the data equipment to put characters into groups and send them. By grouping the data, block synchronization can be used. In long blocks of data, the use of synchronous transmission can save line time.

Timing Must Be Controlled

The timing between the data and the system must be closely controlled in synchronous communications. To overcome timing differences, the modem will control the clocking of data. The modem actually sends the clock signal to the local system, telling it when to transfer bits to the modulator. The receiving modem will watch the phase shifts and know that the occurrences signal bit groups and it will resynchronize the receiving clock. In this manner, the modems and system maintain their clocks in synchronization.

The Bisynchronous communications method (Bisync) is an IBM-designed protocol. The basic protocol is not much different than the point-to-point method. The Bisynchronous communications method has many variations, which include a multipoint protocol, dial-up protocol, the point-to-point protocol, and the transparent.

So far we have been concerned with communications protocols that use "byte" information for both the data and line control. The data being transmitted always fell into eight-bit groupings. The byte protocols are the easiest to handle and program, but they don't really allow the mixing of data types. If we were to use the Bisync as an example, we could not send five-bit data within the block because it might appear in strings as control characters. Also, the five-bit data could not be unpacked by the hardware because it is expecting eight-bit characters.

As communications between systems became more complex, the need arose for protocols that were transparent to the data. The new protocols operated at the bit level where individual bits designated the control functions.

Bit Protocols Are Increasing

The bit protocols in use today usually fall under either the ANSI ADCCP (Advanced Data Communications Control Procedure) or the ISO HDLC (High-level Data Link Control). the standards are fairly broad and the various implementations may only use portions of the various procedures. You would have to check with the vendor to see if his protocol could talk to another vendor's equipment. The bit-oriented protocols are becoming more popular with the introduction of more-sophisticated commuications equipment to the industry.

Packet Switching Networks are a digital version of the dial telephone system. Through the packet networks, terminals and computers can establish digital connections for which they are charged by the time and amount of data transmitted. Connection to the packet networks is by several different methods:

Dial-Up--Using the public dial network, a terminal user can connect with a local "node" of the packet network. Once connected, the user can instruct the network to establish a digital connection to a point in the packet network.

Leased Line--A private line can be connected to the packet node and this line then can be used with a front-end processor to provide multiple input ports for a computer system.

X.25 Link--This is a high-speed trunk link that can be used to place calls and transfer data within the network.

In the network, the nodes are actually computer systems that establish the links and move the data between the users. Within the link the connections between users are called virtual circuits because they do not exist as physical connections but simply as routes in computerized tables. Each node computer acts as a store-and-forward switch for the data.

The public packet-switching networks are usually built around their own proprietary architectures. The main differences in the networks lie in how data is moved and what types of devices can be connected. Some examples of differences in packet networks are:

Computer Sets Up the Routes

Centralized Control--The setting up of virtual circuits is controlled from a central computer. When a user connects to the network and requests a virtual connection, the local node sets up the route as per the instructions of a central computer. The central computer can control loading over the major trucks and prepare for fallback. The only drawback in this case is that a failure of the central control can stop any new users from joining the network.

Distributed Control--When a call is requested, the local node communicates directly with the remote node to set up the virtual circuit. Since the control is distributed, the failure of any of the nodes allows the other nodes to continue their operation.

Routing of Data--If the network uses stochastic routing, it means that each packet will find its own route to the fial destination. The packets will be passed on by each node over the route that has the least data in the direction of the final node. This can result in time delays that vary as the network load changes. Some networks set up the data path through the nodes when the call is connected. With this type of routing, the data will always follow the same route and have the same delay.

The important thing to understand is that the basic concept of all packet-switching networks is to provide low-cost data transmission by sharing the use of communications links. The average voice-grade circuit can handle a 9600-bit-per-second modem, although higher speeds can be reached with newer modems. In normal point-to-point use, the 9600-b/s circuit would support one device talking to a computer. If the device were a terminal, it would be a waste of the data bandwidth, since the average typist would only enter seven to nine characters per second (maximum burst).

Multiplexer Solves the Problem

In private networks, the most-likely solution to this problem is to place a multiplexer on the line. By lowering the terminal speeds to 1200 b/s (120 characters per second), eight terminals could share the one 9600-b/s line.

In a packet network, the nodes act as multiplexers. They take advantage of the intermittent (or statistical) form of data between the terminal and the system. The data the operator types is assembled into blocks called packets. The packets are sent over high-speed lines to a remote node and put out through a port to the computer system.

By using computers in the nodes, the packet network shares the high-speed link over many subscribers. As an example, if a 1200-b/s terminal is only putting out an average of 300 b/s link.

Packet networks are being established in most of the industrialized nations. The networks are being interconnected to provide data-switch on a worldwide basis. Through the use of the X.25 protocol, computers will be able to establish connections automatically with other systems in other countries.

The X.25 protocol is an international standard for the connection of computer systems into the packet networks. By connecting one X.25 link between the node and a computer, the system can set up its own virtual circuits.

Prior to the X.25 protocol being available, the only way to connect to a packet network was via a terminal or a Packet Assembler/Disassembler (PAD). These early PADs weren't based on the present international standard (CCITT X.3), but on the protocol of the network's supplier. One of the problems with the PAD was that it required a separate connection into the computer for each of the virtual circuits being set up.

X.25 Allows Direct Connection

The old style of PAD limited the system to receiving only as many calls as there were PAD ports. It was realized that if the PAD link could be brought directly into tye system, the computer software could perform the assembly and disassembly of packets. The CCITT X.25 standard is what allows the computer to be directly connected to a packet network and substitute for the old PAD or other network interfaces. Over the one X.25 link, a computer can have a dialogue with multiple terminals and computers. The limit to the number of connections that can be established is 4,096, but the system would probably be overloaded with data before the maximum nuber woould be reached.

The X.25 protocol is not limited to public packet networks, but also can be used in private networks as a method of establishing connections and transferring data. The standard is highly useful as a private-network tool where different devices and computers need access over a distributed system. The standard is too complex to describe in detail, but its main use is as an interface to the worldwide packet networks, giving computers the ability to set up their own calls and transfer data.

There are two drawbacks to the worldwide telex network. First, the dependency on the five-level code limits the amount of data that can be represented. There are 60 characters plus two shift codes. when sending telexes between countries, the diacritical marks can't be printed, affecting the correct translation of information.

Second, the line speed of 50 bits per second greatly limits the throughput. This limitation was based originally on use of Teletype machines, which were mechanical in decording and printing, thus greatly limiting their speed.

Computerization Improves Speed

The use of computers and "smart" terminals to file telexes has greatly improved speed, although the customers is still charged as though the message were sent at 50. Most of the record carriers are providing interfaces of 300, 1200 or 2400 b/s into their systems.

The problem of code conversion is being overcome by an attempt to adopt the teletext code. Although an agreement has not been reached on the final code, and although the teletext system is in its infancy, it can be expected to develop into a replacement for record or telex systems. The teletext code, being eight-bit, will be useful in the development of international electronic-mail systems using a global communications network.