Asynchronous Serial Transmission
These brief notes have been extracted from the draft of Principles of Computer Hardware (third edition). We describe the asynchronous serial link, the modem, and the RS232 interface between the computer and modem. These notes provide only the essentials of these topics.
Figure 1 shows the waveform corresponding to a single seven-bit character. In an asynchronous serial transmission system the clocks at the transmitter and receiver responsible for dividing the data stream into bits are not synchronized. The output from the transmitter sits at a mark state whenever data is not being transmitted and the line is idle. The term mark belongs to the early days of data transmission and is represented by a -12V in many systems operating over short distances.
Figure 1 Asynchronous serial transmission
In what follows, a bit period is the shortest time for which the line may be in a logical 1 (mark) or a logical 0 (space) state. When the transmitter wishes to transmit a word, it places the line in a 0 state for one bit period. A space is represented by +12V. When the receiver sees this logical 0, called a start bit, it knows that a character is about to follow. The incoming data stream can then be divided into seven bit periods and the data sampled at the center of each bit. The receiver's clock is not synchronized with the transmitter's clock and the bits are not sampled exactly in the center.
After seven data bits have been sent, a parity bit is transmitted to give a measure of error protection. If the receiver finds that the received parity does not match the calculated parity, an error is flagged and the current character rejected. The parity bit is optional and need not be transmitted.
One or two stop bits at a logical 1 level follow the parity bit. The stop bit carries no information and serves only as a spacer between consecutive characters. After the stop bit has been transmitted, a new character may be sent at any time. Asynchronous serial data links are used largely to transmit data in character form.
If the duration of a single bit is T seconds, the length of a character is given by start bit plus seven data bits plus the parity bit plus the stop bit = 10T. Asynchronous transmission is clearly inefficient, since it requires ten data bits to transmit seven bits of useful information. Several formats for asynchronous data transmission are in common use; for example, eight data bits, no parity, one stop bit.
Bit-rate and Baud-rate
The speed at which a serial data link operates is expressed in bits per second and is typically in the range 110 to over 56,600 bps.
Two units of speed are employed in data transmission. Once is bits/per second (bps) and the other Baud (after Baudot, a pioneer in the days of the telegraph). Bit rate defines the rate at which information flows across a data link. Baud rate defines the switching speed of a signal (i.e., the Baud rate indicates how often a signal changes state).
For a binary two-level signal, a data rate of one bit per second is equivalent to one Baud; for example, a modem transmitting binary data at 1,200 bps is said to operate at 1,200 Baud. Suppose a data transmission system uses signals with 16 possible discrete levels. Each signal element can have one of 16 = 24 different values; that is a signal element encodes 4 bits. If the 16-level signals are transmitted at 1,200 Baud, the data rate is 4 x 1,200 = 4,800 bps.
Modulation and Data Transmission
We are now going to look at a topic called modulation, the means of modifying signals to make them suitable for transmission over a particular channel. A bandpass channel like a telephone channel can transmit sine waves within its bandwidth but can't transmit digital pulses. If a sequence of binary signals were presented to one end of a telephone network, the digital signals would be so severely distorted that they would be unrecognizable at the receiving end of the circuit.
Because the telephone network can transmit voice-band signals in the range 300 to 3,300 Hz, various ways of converting digital information into speech-like signals have been investigated. Figure 2 shows how the digital data can be used to change, or modulate, the amplitude of a sine wave in sympathy with a digital signal. This technique is known as amplitude modulation or AM. The equipment needed to generate such a signal is called a modulator and that needed to extract the digital data from the resulting signal is called a demodulator. The interface between a computer and a telephone system is called a MODEM (modulator-demodulator). Because AM is more sensitive to noise (i.e., interference) than other modulation techniques, it is not widely used in data transmission.
Figure 2 Amplitude modulation
Instead of modulating a sine wave by changing its amplitude, it's possible to change its frequency in sympathy with the digital data. In a binary system, one frequency represents one binary value and a different frequency represents the other. Figure 3 shows a frequency modulated (FM) signal. FM is widely used because it has a better tolerance to noise than AM (i.e., it is less affected by various forms of interference).
Figure 3 Frequency modulation
Figure 4 illustrates another form of modulation called phase modulation (PM). In this case, the phase of the sine wave is changed in sympathy with the digital signal. PM is widely used and has fairly similar characteristics to FM. If the phase change corresponding to a logical 1 is 180° , and 0° (no change) corresponds to a logical 0, one bit of information can be transmitted in each time slot (Figure 4). If, however, the phase is shifted by multiples of 90° , two bits at a time can be transmitted (Figure 5).
Figure 4 Phase modulation
Figure 5 Differential phase modulation
High-speed Modems
Modems operate over a wide range of bit rates. Until the mid 1990s most modems operated between 300 bps to 9,600 bps. Low bit rates were associated with the switched telephone network where some lines were very poor and signal impairments reduced the data rate to 2,400 bps or below. The higher rates of 4,800 bps and 9,600 bps were generally found on privately leased lines where the telephone company offered a higher grade of service.
The growth of the Internet provided a mass market for high-speed modems. Improved modulation techniques and better signal-processing technology has a massive impact on modem design. By the mid 90s, low cost modems operated at 14.4K baud or 28.8K baud. By 1998, modems capable of operating at 56K baud over conventional telephone lines were available for the price of a 1200 bps modem only a decade earlier.
High-speed modems operate by simultaneously changing the amplitude and phase of a signal. This modulation technique is called quadrature amplitude modulation (QAM). A QAM signal can be represented mathematically by the expression S x sin(wt) + C x cos(wt), where S and C are two constants. The term "quadrature" is used because a sine wave and a cosine wave of the same frequency and amplitude are almost identical. The only difference is that a sine wave and a cosine wave are 90° out of phase (90° represents ¼ of 360° —hence quadrature). Figure 6 demonstrates a 32-point QAM constellation in which each point represents one of 32 discrete signals. A signal element encodes a 5-bit value which means a modem with a signaling speed of 2400 baud can transmit at 12,000 bps.
Figure 6 The 32-point QAM constellation
High-speed Transmission over the PSTN
The backbone of the POTS (plain old telephone system) is anything but plain. Data can be transmitted across the world via satellite, terrestrial microwave links, and fiber optic links at very high rates. The factor that limits the rate at which data can be transmitted is known as "the last mile"; that is, the connection between your phone and the global network at your local switching center.
ISDN
A technology called ISDN (integrated services digital network) was developed in the 1980s to help overcome the bandwidth limitations imposed by the last mile. ISDN was developed largely for professional and business applications and is now available to anyone with a personal computer. There are two variants of ISDN—basic rate services and primary rate services. The basic rate service is intended for small businesses and provides three fully duplex channels. Two of these so-called B channels can carry voice or data and the third D channel is used to carry control information. B channels operate at 64K bps and the D channel at 16K bps.
ISDN's popularity is due to its relatively low cost and the high quality of service it offers over the telephone line. You can combine the two B channels to achieve a data rate of 128K bps. You can even use the D control channel (simultaneously) to provides an auxiliary channel at 9.6K bps. Note that ISDN can handle both voice and data transmission simultaneously.
Several protocols have been designed to control ISDN systems. V.110 and V.120 are used to connect an ISDN communications devices to high-speed ISDN lines. ISDN took long time from its first implementation to its adoption by many businesses. However, newer technologies have been devised to overcome the last mile problem and ISDN will probably never become as commonplace as some had anticipated.
ADSL
If there's one thing you can guarantee in the computing world, it's that yesterday's state-of-the-art technology becomes the current standard, a new state-of-the-art technology is emerging. Just as ISDN was becoming popular in the late 1990s, as system called ADSL (asymmetric digital subscriber line) was being developed as a new high-speed "last mile" system.
As we've said, telephone lines have a bandwidth of 3000 Hz that limits the maximum rate at which data can be transmitted. In fact, the twisted wire pair between your home and the telephone company has a much higher bandwidth. The bandwidth of a typical twisted pair less than about 3 miles is over 1 MHz.
Asymmetric Digital Subscriber Line technology exploits the available bandwidth of the local connection. The bandwidth of the telephone link is divided into a number of 4 KHz slices as figure 7 demonstrates. The first slice from 0 to 4 KHz represents the conventional telephone bandwidth. Frequencies between 4 kHz and 24 KHz aren't used in order to provide a guard band to stop the higher frequencies interfering with conventional telephone equipment.
The spectrum between 24 kHz and 1.1 MHz is divided into 249 separate 4 KHz channels in the same way as the FM band is divided into slots for the various broadcasting stations. A data signal can be assigned one of these slices and its spectrum tailored to fit its allocated 4 KHz slot. At the other end of the link, the signal in that 4 KHz slot is converted back into the data signal. Until recently it was very difficult to perform these operations. The advent of low-cost digital signal processing has made it much easier to process signals (i.e., to shift their range of frequencies from one band to another).
The characteristics of these slots vary with frequency; for example, there is much more attenuation of signals in slots close to 1.1 MHz. The terminal equipment is able to use the better channels to carry high data rates and to allocate the higher frequency channels to slower bit rates.
Figure 7 Dividing a 1.1 MHz bandwidth into 4 kHz slots
The RS232C Interface
The first really universal standard for the physical connection between computer and modem was published in 1969 by the Electronic Industry Association (EIA) in the USA and is known as RS232C (Recommended Standard 232 version C). Since then the standard has been revised (e.g., RS232D and RS232E).
RS232 specifies the plug and socket at the modem and the digital equipment (i.e., their mechanics), the nature of the transmission path and the signals required to control the operation of the modem (i.e., the functionality of the data link).
From the point of view of the standard, the modem is known as data communications equipment (DCE) and the digital equipment to be connected to the modem is known as data terminal equipment (DTE). Figure 8 illustrates the role played by the RS232 standard in linking DCE to DTE.
Figure 8 Linking DTE to DCE with the RS232 data link
Because RS232 was intended for DTE to DCE links, its functions are very largely those needed to control a modem.
RS232C Control Lines
The RS232 standard describes the functions carried out by several control signals between the DTE and the DCE. The following control signals implement most of the important functions of an R232 DTE to DCE link.
Request to send (RTS) This is a signal from the DTE to the DCE. When asserted, RTS indicates to the DCE that the DTE wishes to transmit data to it.
Clear to send (CTS) This is a signal from the DCE to the DTE and, when asserted, indicates that the DCE is ready to receive data from the DTE.
Data set ready (DSR) This is a signal from the DCE to the DTE which indicates the readiness of the DCE. When this signal is asserted, the DCE is able to receive from the DTE. DSR indicates that the DCE (usually a modem) is switched on and is in its normal functioning mode (as opposed to its self-test mode).
Data terminal ready (DTR) This is a signal from the DTE to the DCE. When asserted, DTR indicates that the DTE is ready to accept data from the DCE. In systems with a modem, it maintains the connection and keeps the channel open. If DTR is negated, the communication path is broken. In everyday terms, negating DTR is the same as hanging up a phone.
Example
How long does it take a computer to transmit a certain picture to a remote site over the telephone system, given the following data?
1. The image measures 4 inches by 2 inches.
2. The image has been scanned at a resolution of 200 pixels/inch.
3. Each pixel represents a 32-level grey-scale value (i.e., 32 steps from white to black).
4. The data is transmitted asynchronously with one start bit, eight data bits, no parity bit, and one stop bit.
5. The signalling speed of the modem is 2,400 baud.
6. The modem uses 256-point QAM to modulate the signal.
Note: A pixel is a picture element and corresponds to a "dot". A pixel can have attributes such as colour.
Solution
a. The total number of pixels is:
horizontal pixels x vertical pixels = (4 x 200) x (2 x 200) = 800 x 400 = 320,000 pixels
b. Each pixel represents one of 32 levels of grey. Therefore, a pixel is encoded as 5 bits (25 = 32).
c. The total number of bits to be transmitted is:
pixels x bits/pixel = 320,000 x 5 = 1,600,000 bits/image.
d. The switching (signalling) speed is 2400 baud and each signal is 1 of 256 different values. That is, each signal carries 8 bits (because 28 = 256).
e. The transmitted bit-rate is given by baud rate x bits/signal = 2,400 x 8 = 19,200 bits/s
f. Each unit of data transmitted (i.e., each character) consists of 8 data bits in a frame consisting of 1 start bit + 8 data bits + 0 parity bit + 1 stop bit. It takes 10 bits in a frame to transmit 8 data bits. The effective data transmission rate is therefore reduced by 8/10. Consequently, the modem transmits at
19,200 x 8/10 = 15,360 bits/s.
g. The time taken to transmit the image is (total bits)/(transmitted bit rate) = 1,600,000/15,360 = 104 s.
h. In practice, the value would be higher to account for any time between successive characters and the overhead needed to set up the call and to deal with its progress.
i. Note that many real data transmission systems first compress the data rather than sending the full 1,600,000 bits. Since much of most images has a constant intensity (e.g., most of a printed page is white), data can be run length encoded. That is, you transmit the number of pixels in a run of constant intensity.