Overview
This chapter
examines general data communications principles, and how they apply to building
a network. Hardware characteristics are also discussed. The network we use today
is part analog and part digital and the struggle is how to best transmit data.
Using a voice network has many advantages, but some disadvantages too. A voice
session has so much redundancy that a few audible noise bursts have little effect
on the message. However, a few missed bits in a banking transaction are critical.
The data network was not developed by one company as the telephone network was
so there is always the problem of proprietary interfaces. Many standards have
been developed to simplify data communications. A telecommunications manager
should consider several factors as he looks at data communications equipment:
·
compatibility
with standards
·
compatibility
with existing equipment
·
support of
the manufacturer and its representatives
·
compatibility
with the network management systems
The basic information element used by the computer is called a bit. The bit is the smallest element in the binary system and is either a 1 or a 0. These two digits are represented by two different voltage states in the DTE. A group of eight bits is called a byte or octet. Bytes travel over parallel paths within the computer. However, parallel transmission is limited to a few feet, so these bytes are converted into a serial stream.
Code compatibility
is essential. Most intelligent terminals can be programmed for code conversion.
The types of codes are:
·
Baudot
early teletypewriters used 5 level code which allows 2 to the 5th
power or 32 characters
·
ASCII
American Standard Code for Information Interchange allows 2 to the 7th
power or 128 combinations. Actually 8 bits are sent with one being used for
error detection.
·
EBCDIC
Extended Binary Coded Decimal Interchange Code IBMs code uses the full
8 bits for data, allowing 256 characters.
Speeds in
data communication are measured in bits/second. Backbone circuits on the Internet
run at 10 Gb/s. Voice grade lines can, theoretically, support up to 56 kb/s.
Two terms express capacity:
·
Bit rate
number of bits per second the channel can carry
·
Baud rate number of cycles or symbols
per second the channel can handle
If the signal is encoded at 1 bit/cycle, then baud rate = bit rate. However, encoding techniques allow more than one bit/cycle to be used so 2400 baud rate could be 19.2 kb/s if 8 bits were encoded in one cycle..
If the signal
is to be sent over a voice-grade channel, the modem must convert the signal
to a combination of analog tone and amplitude and phase changes.
·
Analog modulation least used method alone because
it is susceptible to noise, but often used in conjunction with the others
·
Frequency modulation inexpensive for low speed modems
·
Phase-shift modulation used for speeds greater
than 300 b/s
High-speed models use QAM to send multiple bits per hertz. If an analog channel is limited to 2400 baud then to reach 9600 b/s, 4 bits/Hz must be encoded. The resulting 2 to the 4th encoding yields 16 combinations that each symbol can represent. In QAM, two-carrier tones combine to produce the modems output signal. Any combination of the 4 bits can be encoded into an X-Y plot, which represent phase and amplitude combination. This two- dimension diagram is called a signal constellation. Even higher rates can be modulated with each additional bit doubling the number of signal points. If noise or jitter occurs, the received point will be displaced from its ideal location, so the receiver will make a best guess about what plot point was sent.
Trellis-Coded Modulation
A reliable method of encoding data signals. In a 14,400 b/s modem, data is presented to a TCM modulator in 6-bit groups. Two bits are separated from the other 4 bits and 1 code bit is added. The result is two groups a 3-bit and a 4-bit group. The result is a 2 to the 7th, or 128 point signal constellation. However, only 6 bits are valid so only 64 points are needed to transmit the signal and only certain patterns of the signal points are termed valid. If there are line impairments, the decoder selects the most likely valid sequence.
There are
two modes of transmission:
·
Full-duplex data
systems transmit data in both
directions simultaneously use four-wire circuit of split channel modem on
two wire circuit
·
Half-duplex data systems transmit
data in only one direction at a time the channel reverses for transmission
in the other direction
Split channel modems provide the equivalent of four-wire operation by dividing the voice channel into two segments, one for transmit and one for receive. This effectively divides the bandwidth of the channel by two.
All data communications channels require synchronization to keep the sending and receiving ends in step.
· Asynchronous also called start/stop; starts with a zero(or space) level bit followed by data bits and ending with a 1(or mark) level stop bit; transmitted in character mode; chief drawback is the extra two bits or overhead that carry no information
· Synchronous sent in block mode with information characters sandwiched between header and trailer records; less overhead; requires a clock signal that the modem extracts from the incoming bit stream to keep in synch; advantages include: greater throughput, ability to use sophisticated error-correction techniques disadvantage include: complexity, lack of standardization, need for protocol compatibility and intelligence in the DTE
Errors occur in all data communications. Transmissions such as banking transactions require complete accuracy.
Any transmission
medium using analog modulation techniques is subject to external noise that
can affect the amplitude of the signal. These can include:
·
atmospheric
condition like lightning
·
switching
to standby channels cause a momentary delay
·
changes
in the phase of the received signal cause by instability in carrier supplies
The simplest way to detect errors is parity checking (also called vertical redundancy checking (VRC)), a technique used on asynchronous circuits. The eighth bit in ASCII code is used for parity and is set at a one or zero depending on the number of 1s or 0s in the 7-bit data word. The determination of whether the parity is odd, even, or zero is set by the network. This only indicates that there was an error, but does nothing about it. In addition, data is coming at such high speeds that this is not as useful as it once was and is now turned off by most machines.
Echo Checking
Over full-duplex circuits, errors can be detected by programming the receiving device to echo the received characters to the ending end called echo checking. Disadvantages include: the error could occur on the way back; high transmission speed makes it impossible to read with reliability; identifies errors but doesnt correct them. Most dial-up modems now have a built-in error correction using the V.42 error correction standard.
Cyclical Redundancy Checking
Most synchronous data networks use cyclical redundancy checking (CRC). The characters in the data frame are processed against a complex polynomial that always produces a remainder. This 16-bit remainder is transmitted following the data block. The receiving end processes the frame against the same polynomial and compares the received and computed remainders. If they do not agree, the entire data block is retransmitted. This is such effective data detection and correction that the transmission is considered error free. The Block error rate (BLER) is calculated by the dividing the number of blocks in error by the total number of data blocks transmitted and is the measure of quality of the data link.
Forward Error Checking
If the BLER is too great, the throughput (defined to be the number of information bits correctly transferred) may be reduced to an unacceptable level. Using forward error checking (FEC) can reduce the error rate. The encoder on the transmitting side generates redundant code bits so that the signal includes both signal and redundant bits. On the receiving end, the redundant bits are regenerated and compared to the redundant bits that are sent. If they do not match, the receiving end uses the redundant bits to generate the most likely bit combination.
This is
the measure of information bits that are correctly transferred. The factors
that limit throughput are:
·
Modem
speed - voice channels transmit up to 56kb/s
·
Circuit
error rate higher the rate, the lower the throughput
·
Half-
or full-duplex - generally, full-duplex is faster
·
Protocol
protocol dictates the number of overhead bits and error correction methodology
·
Overhead
bits
·
Size
of the individual data block
·
Speed
of the transmisison medium
A facility
is the generic term used to describe the combination of local loops and long-haul
circuits that support communications. Types include:
·
Interconnected
local area networks
·
Automatic
teller machines
·
Credit
card verification
·
Single
host that supports multi-location terminals
·
Point-of-sale
terminals
·
Electronic
mail
·
Surfing
the World Wide Web
There are
several ways that networks can be designed. They are:
·
Point-to-Point Circuits directly wired between the stations
on the network; cost effective for high-speed communication; expensive for keyboard
applications.
·
Multidrop Circuits
- a host computer sends
a polling message to each station in turn and if the station has traffic, it
sends it or a negative response; significant overhead consumed by the polling
messages and negative responses.
·
Circuit Switching
central switch connected to stations in a star configuration; communications
is between the stations and the switch; circuit is not fully utilized.
·
Message Switching
- also
called store and forward; stations look for computers that will accept messages
and store them for later delivery.
·
Packet Switching
- have
control nodes that are interconnected by trunks; the nodes check for errors,
then pass the packets on to the next node through virtual circuits and finally
to its destination
·
Frame Relay - similar to the packet switch
but none of the error detection; used over high-quality fiber optics; relies
on the end computer to check for errors.
·
Cell Relay - combination multiplexing and
switching protocol; nodes are high-speed switching devices instead of store
and forward devices; used in voice and data across broadband networks.
There are
three types of terminals to be used on a network:
·
Dumb Terminals have no processing power; not
addressable; cannot respond to polling messages; no error detection; located
near host computer
·
Smart
Terminals nonprogrammable; addressable; data stored in buffer to transmit
block mode; limited processing capabilities
·
Intelligent
Terminals contains own processor; runs applications; PC is example;
provides better line utilization; communication software must run in the terminal
·
Terminal
Emulation can emulate any of the 3 terminal types
A modem is used to connect to a computer/terminal to the network. The V.90 standard, developed in 1988, allows up to 56kb/s transmission. There are three major types of modems:
· Dial-Up Modems become a commodity; either plug into computer slot or are self-contained; support the V.42 error-connection protocol; function in full-duplex mode; include these features:
· Private Line Modems being replaced by digital counterparts; different manufacturers use proprietary formats
· Special-Purpose Modems these include:
When a telecommunications
manager evaluates modems for purchase, he should consider the following:
·
Dial-up
versus private line do you need to handle both dial-up and private lines?
If so, this is more expensive.
·
Standards
compatibility does the modem adhere to ITU standards which provide error correction
and compression?
·
Modem
reversal time if you have a half-duplex line, how important is the speed with
which the modem reverses from send to receive? What is the cost?
·
Modulation
method what is the modulation scheme of the modem, does it fit a standard,
and how important is the speed versus cost?
·
Speed
what is the lowest acceptable transmission speed, since cost increases with
speed?
·
Operating
mode are you interested in full- or half- duplex mode?
·
Equalization
method do you need fixed or adaptive equalization? Will the modem have to
be able to work over an unconditioned line? Cost?
·
Diagnostic
capability how much information do you need to be displayed for network management
diagnostics?
Data Service Unit/Channel Service Unit (DSU/CSU)
A DSU/CSU connects the DTE to a digital circuit, in order to convert a unipolar signal to a bipolar signal for transmission on a digital circuit.
Multiplexers and Concentrators
In order
to make use of idle capacity on a data circuit, multiplexers collect data from
multiple stations and combine them into a single high-speed bit stream. Multiplexers
provide end-to-end error checking and correction, and circuit sharing. There
are two types of multiplexers:
·
Time division
multiplexers (TDMs) each station is assigned a time slot and the TDM collects
data from each station in turn
·
Statistical
multiplexers (statmux) each pair of stations is assigned a time slot
and the statmux collects data from the DTE
Features
of the multiplexers include:
·
alternate routing
can transmit data around network congestion
·
terminal-to-host
mapping can determine route for user to log onto network and address any host
·
network management
capability remote network management system can monitor the network through
an interface into the multiplexer
·
integrated
CSU/DSU eliminates the need for a separate outboard device
A concentrator is similar to the multiplexer except that is usually a single-end device. At the terminal end, devices connect to the concentrator that connects to the facility, and at the host end, the facility is routed directly into the front-end processor.
When evaluating
multiplexers/concentrators, one should consider the following:
·
Line speed
what speed can the line handle and is the multiplexer upgradable to higher
speeds later with a simple card change?
·
Number and
speed of ports supported how many ports can the multiplexer handle and what
is the port speed ?
·
Redundancy
does it have redundant power supplies and processors?
·
Protocol support
which protocols do you need to support? Do you need to support only asynchronous
or do you need to support X.25 connections?
·
Security
how many levels of security support do you need?
·
Network Management
Capability do you need to use the multiplexer to support the network management
protocol?
Packet networks are robust systems because of the alternate route capability. They are interconnected in a mesh configuration that allows at least two alternative paths for data to be transmitted. At each note of the path, the data is checked for errors. A PAD (packet assembler/disassembler) creates the packet at the sending end and breaks it apart at the receiving end. The interface between the PAD and the distribution node called X.25. This system that checks each packet for errors is being overtaken by the same technology without the error checking called frame relay access.
When the
telecommunications manager is evaluating packet switching equipment, he looks
at the same issues as defined above for multiplexers as well as:
·
Absolute
delay what is the delay through the packet switching equipment?
·
Access
method do you need a dedicated circuit to the vendors PAD, or can a dial-up
circuit work? If you have your own PAD, do you have an X.25 circuit from the
premises to the node?
There are
several pieces of ancillary equipment that facilitate using unique application:
·
Protocol Converters
connect incompatible devices;
gateway is an example
·
Dial Backup Units
provides continuity of service
if the dedicated line fails
·
Multidrop
Bridges creates multi-drop circuits by bridging several point-to-point
lines