Chapter 4
Local Area
in the
Internet Age
by Alan Dennis
Copyright © 2002 John Wiley & Sons, Inc.
Copyright John Wiley & Sons, Inc. All rights reserved.
Reproduction or translation of this work beyond that named in
Section 117 of the United States Copyright Act without the
express written consent of the copyright owner is unlawful.
Requests for further information should be addressed to the
Permissions Department, John Wiley & Sons, Inc. Adopters of
the textbook are granted permission to make back-up copies for
their own use only, to make copies for distribution to students of
the course the textbook is used in, and to modify this material to
best suit their instructional needs. Under no circumstances can
copies be made for resale. The Publisher assumes no
responsibility for errors, omissions, or damages, caused by the
use of these programs or from the use of the information
contained herein.
Chapter 4. Learning Objectives
• Understand the major components of LANs
• Understand shared Ethernet and switched Ethernet
• Understand Ethernet media access control
• Understand Ethernet error control
• Be familiar with Ethernet message delineation
• Be familiar with how data is transmitted through
physical circuits
• Understand the best practice recommendations for
LAN design
Chapter 4. Outline
• Introduction
• Topology
– Shared Ethernet, Switched Ethernet
• Media Access Control
• Error Control
– Error Detection, Error Correction
• Message Delineation
– Codes, Frame Layout, Frame Size
• Data Transmission in the Physical Layer
– 10BaseT, 100BaseT and 100BaseF, 1GbE, 10GbE
• The Best Practice LAN
– Effective Data Rates, Costs, Recommendations
• Over 95% of LANs in use today use the Ethernet
protocol standardized as IEEE 802.3.
• LANs generally have two hardware layers: the
data link layer and the physical layer.
• The data link layer performs three basic functions:
message delineation, medium access control and
error control.
• Ethernet can detect and correct errors, but since
errors are rare in modern LANs, error correction is
• 10BaseT and 100BaseT are example of physical
layer Ethernet protocols (10 stands for 10 Mbps,
T stands for twisted pair).
• Topology refers to the geometric layout of the
• A logical topology is how the network works
conceptually, rather like the logical DFD in
systems analysis.
• Physical topology refers to how the network is
physically connected.
• Ethernet’s two forms shared and switched
Ethernet, use bus and star logical topologies,
Shared Ethernet (Figure 4-1)
• In addition to the computers, shared Ethernet
LANs contains three basic components:
– Network Interface Cards
– Cables
– Hubs
Figure 4-1 Local area network components
Network Interface Cards
• Network interface cards, also called network cards
and network adapters include a cable socket
allowing computers to be connected to the network.
• NICs are part of both the physical and data link layer
and include a unique data link layer address
(sometimes called a MAC address), placed in them
by their manufacturer.
• Before sending data onto the network, the network
card also organizes data into frames and then sends
them out on the network.
• Notebook computers often use NICs that are
plugged into the PCMCIA port.
• Each computer is physically connected to
the network using a cable.
• The cables used on Ethernet LANs are
either twisted-pair or optical fiber cables.
• Data can flow through cables in one of three
modes (Figure 4-2):
– One way only (simplex)
– Both ways, one way at a time (half-duplex)
– Both ways at the same time (full-duplex)
Figure 4-2 Simplex, half-duplex, and full-duplex
Hubs (Figure 4-3)
• Hubs act as junction boxes, linking cables from
several computers on a network. Hubs are usually
sold with 4, 8, 16 or 24 ports.
• Some hubs allow connection of more than one
kind of cabling, such as UTP and coax.
• Hubs also repeat (reconstruct and strengthen)
incoming signals. This is important since all
signals become weaker with distance.
• The maximum LAN segment distance for a cable
can therefore be extended using hubs.
Figure 4-3 Network hub
Shared Ethernet Topology (Figure 4-4)
• Shared Ethernet’s logical topology is a bus topology.
• This means all computers on the network receive
messages from all other computers, whether the
message is intended for those computers or not.
• When a frame is received by a computer, the first
task is to read the frame’s destination address to see
if the message is meant for it or not.
• Ethernets today use a physical star topology, with
the network’s computers linked into hubs.
• It is also common to link use multiple hubs to form
more complex physical topologies, enabling the
networks to span longer distances (Figure 4-5).
Figure 4-4 Ethernet topology
Figure 4-5 An example of an Ethernet local
area network with two hubs
Switched Ethernet Topology
• Switched Ethernet uses switches instead of hubs.
• Switches that make switching decisions based on
data link layer addresses are called workgroup or
layer-2 switches.
• While a hub broadcasts frames to all ports, the
switch reads the destination address of the frame
and only sends it to the corresponding port.
• The effect is to turn the network into a group of
point-to-point circuits and to change the logical
topology of the network from a bus to a star
(Figure 4-6).
Figure 4-6 Ethernet topology
Basic Switch Operation
• Switches have onboard chips that make forwarding decisions
using forwarding tables (similar to routing tables).
• When a frame is received, the switch reads its [data link
layer] destination address and sends the frame out the
corresponding port in its forwarding table.
• When a switch is first turned on, its forwarding table is
empty. Switches learn which ports correspond to which
computers by reading the source addresses of the incoming
frames and noting the port number the frame arrived on.
• If the switch’s forwarding table does not have the destination
address of the frame, it broadcasts the frame to all ports.
• Thus, a switch starts by working like a hub and then works
more and more as a switch as it fills its forwarding table.
Managing LAN traffic using switches
• Switch differ from hubs in two important ways.
• First, unlike a hub, which broadcasts incoming
frames out all the ports of the hub, a switch only
sends an incoming frames out the port corresponding
to its destination computer. This greatly reduces
network traffic and prevents most collisions
• Second, switches use a store-and-forward approach
to managing LAN traffic. If two frames arrive at a
switch at the same time, the second frame is
temporarily stored in memory until the switch is
done processing the first frame.
Figure 4-7 802.3 Ethernet versus switched Ethernet
Media Access Control
Media Access Control
• Ethernet’s medium access control protocol,
CSMA/CD, is contention-based.
• This means two devices can transmit at the same
time. When they do, their frames collide, become
garbled and need to be retransmitted.
• Stands for: Carrier Sense Multiple Access w/
Collision Detect
– Carrier Sense: devices listen to see if another the
network is busy before transmitting.
– Multiple Access: multiple devices have network access.
– Collision Detect: if a collision is detected, the devices
sending the frames wait a random amount of time before
resending the frame (It has to be random in order to avoid
another collision).
Error Control
Error Control
• Data communications are prone to errors,
such as noise spikes from power surges,
crosstalk from wires being too close
together, and echoes due to faulty
• A common way to prevent errors is to use
shielded cables.
• Error control combines the techniques for
detecting errors with those for correcting
Error Detection with Ethernet
• Ethernet’s error detection scheme is called Cyclic
Redundancy Check (CRC).
• CRC is computed by treating a message as one
long binary number, P.
• Before transmission, the data link layer hardware
divides P by a fixed binary number, G, resulting in
a whole number, Q, and a remainder, R/G.
• The remainder, R, is written into the outgoing
frame and then recalculated by the receiver.
• CRC-16 (99.998% effective for errors greater than
16 bits in length) and CRC-32 (99.99999998%for
errors greater than 32 bits long) are in common
use today.
Error Correction via Retransmission
• The default setting for Ethernet is to use error
detection without error correction. Bad frames are
simply discarded.
• When error correction is used, however, errors are
corrected by retransmitting the frames found to be
in error.
• The process of requesting a frame be resent is
called an Automatic Repeat Request or ARQ.
• When error correction is turned on, Ethernet uses
stop-and-wait ARQ.
Stop-and-Wait ARQ (Figure 4-8)
• With stop-and-wait ARQ, the sender first sends a
packet, then waits to hear from the receiver.
• If the packet is received without error, an
acknowledgement (ACK) is sent back by the
receiver and the next packet is sent.
• If the receiver detects an error in the packet that
was just sent, the receiver sends back a negative
acknowledgement (NAK) and the sender resends
the packet again.
Figure 4-8 Stop and wait ARQ
Message Delineation
• Written languages use symbols, but computers send
signals in 1s and 0s (bits).
• Each written character needs a bit code in order to be
used by a computer. A set of these codes for a
language is called a coding scheme.
• The mostly commonly used coding scheme is ASCII:
the American Standard Code for Information
Interchange, originally used a 7-bit code (27 = 128
• An 8-bit version is now in use (an 8-bit coding
scheme has 28 = 256 combinations).
• Ethernet usually uses 8-bit ASCII.
Frame Layout (Figure 4-9)
• The latest version of Ethernet frame format, defined
by the IEEE 802.3ac standard, consists of four parts:
preamble, MAC header, LLC protocol data unit and
the MAC trailer.
• The frame begins with an 8-byte preamble. When the
receiver sees the preamble’s bit pattern, it knows a
frame is beginning. The 8th byte is the start of frame
delimiter (10101011).
• The last section of the frame is the 4-byte frame
check sequence (FCS). This is where the CRC-32
value for the frame is stored.
MAC Header (Figure 4-9)
• The section following the preamble in the Ethernet
frame is the MAC header.
• The first two fields of the MAC header are the 6-byte
source and destination addresses.
• The last field is the 2-byte field length, which gives
the length of the data field. Because Ethernet has
variable length data fields, their length needs to be
known. Without it, the receiver will not know where
the frame ends.
• The third field is the 4-byte virtual LAN (VLAN) tag.
When in use, the first two bytes are set to 81-00 (an
impossible field length) which tells the receiver that
the VLAN tag field is being used.
LLC Protocol Data Unit (Fig. 4-9)
• The LLC protocol data unit follows the MAC header.
• The first two fields are the Destination Service Access
Point (DSAP) and the Source Service Access Point
(SSAP), which function like the port number in TCP;
i.e., telling the internetwork layer which software to
use to process the frame.
• The control field holds frame sequence numbers and
ACKs and NAKs and is used for error control.
• The data field contains the message being
transmitted. It is typically an IP packet. It must be at
least 43 bytes long and must not be more than 1497
bytes in length.
Figure 4-9 Ethernet 802.3ac frame layout
Data Transmission In The
Physical Layer
Analog and Digital Data
• A fundamental physical layer distinction is between
digital and analog forms of data.
• Computers produce digital data such a binary code
which is either on or off or a zero or one.
• Telephones produce continuously varying electrical
signals. This is an example of analog data.
• Most computers transmit data in digital form, but can
convert it to analog form by using a modem.
Transmission Modes
• Data can also be sent either in serial or in
parallel modes:
• Parallel mode (Figure 4-10a): uses several
wires, each wire sending one bit at the same
time as the other wires in its cable.
– A parallel printer cable sends 8 bits together.
– Your computer’s processor and motherboard also use
parallel busses to move data around.
• Serial Mode (Figure 4-10b): sends one bit after
another over a single line. Serial mode is slower
than parallel, but can be used over longer
distances because the bits stay in the order they
were sent, while bits sent in parallel mode tend
to spread out over long distances.
Figure 4-10a Serial transmission of an 8 bit code
Figure 4-10b Parallel transmission of an 8 bit code
Ethernet Physical Layer Standards
• The three most commonly used forms of
Ethernet today are:
• 10BaseT (10 = 10Mbps, T = twisted pair
and base = baseband, meaning one channel)
• Fast Ethernet which includes 100BaseT
and 100BaseF (F = fiber)
• Gigabit Ethernet which includes 1 Gigabit
Ethernet (1 GbE) and 10 GbE (10 Gbps).
Even faster versions are being developed.
The 10BaseT Ethernet Standard
• 10BaseT uses twisted-pair cable.
• By far the most common form today is using 8wire, category 5 cables with RJ-45 connectors
(similar to phone jacks but larger).
• Twisted pair cables can be either unshielded
(UTP) or shielded (STP).
• Shielding reduces interference from outside
sources that may cause transmission errors.
• The maximum possible cable length is 100 meters.
10BaseT Data Transmission (Figure 4-13)
• To transmit at 10Mbps, 10BaseT uses a 10 MHz
signaling rate, sending a bit every 100 nanoseconds.
• To preserve the synchronization of the signal between
sender and receiver, 10BaseT uses Manchester
encoding in which the bit value is defined by a midbit transition.
• The data is not defined in terms of whether the signal
is at a high or low voltage, but by whether the mid-bit
transition goes from a high to a low value (binary 0)
or from a low to a high value (defining a binary 1).
• Figure 4-13 shows a an example of Manchester
Figure 4-13 Manchester encoding
Fast Ethernet: 100BaseT
• There are two forms of fast Ethernet, 100BaseT and
100BaseF. 100BaseT uses twisted-pair cable.
• Signals are sent using a combination of 4B5B coding
and MLT-3 transmission.
• With 4B5B coding, five bits are sent together, four
data bits followed by a fifth synchronization bit.
• Multi-Level Transmission 3 uses three levels of
voltage to send these bits, +1, 0, and –1. Data is
encoded by changing the voltage to the next level.
• To transmit a binary 1, the voltage is changed to an
adjacent level, such as from –1 to 0 or from 0 to +1.
• 100BaseF uses fiber optic cable. Light
created by an LED (light-emitting diode) or
laser is sent down a thin glass or plastic fiber.
• Fiber optic cable structure (from center):
– Core (v. small, 5-50 microns, ~ the size of
a single hair)
– Cladding, that reflects the signal back into
the core
– Protective outer jacket
Fiber Optic Cable Types
• Types of Optical Fiber (Figure 4-14):
• Multimode (MMF) is cheap, but the signal spreads out over
short distances (up to ~500m).
• Graded index MMF reduces the spreading problem by
changing the refractive properties of the fiber to periodically
refocus the signal. Can be used over distances of up to about
1000 meters.
• Single mode is expensive because it is more difficult to
manufacture, but the signal can be sent over longer distances
(up to 100 kilometers) without spreading.
• Three most commonly used fiber types for LANs are:
62.5/125 MMF, 50/125 MMF and 10/125 SMF, (the numbers
are the core and cladding diameters, respectively).
100BaseF Signaling
• 100BaseF uses a pair of 62.5/125 MMF fibers, one
strand for sending, one for receiving.
• Almost always used with switched topologies.
• Maximum segment length (distance from computer
to hub or switch) is 412 meters.
• Transmits using a bright light to indicate a 1 and a
dim light to indicate a zero.
• This is called non-return to zero, since the signal
never goes completely off.
Figure 4-14 Fiber-optic cable
Gigabit Ethernet (1GbE)
• Gigabit Ethernet is the newest family of Ethernet
protocols running at speeds of 1 Gbps and above.
• Can use either twisted pair or fiber.
• 1000BaseT uses cat-5 UTP, using all four wire
pairs operating in parallel.
• Gigabit Ethernet uses the same 125 MHz clock as
100BaseT. It is able to achieve the 1Gbps data rate
by sending data over four lines simultaneously.
This enables data to be sent at a rate of 500Mbps.
• 1000Mbps is achieved by sending 2 bits per
interval. The technique used by 1000BaseT for
this is called PAM-5.
PAM-5 encoding (Figure 4-15)
• 100BaseT encodes data using a technique called
pulse amplitude modulation 5.
• PAM-5 uses 5 different voltage levels, four to send
data and one as a control bit. The data bits are
defined as follows:
00  - 1.0 volts
01  - 0.5 volts
10  +0.5 volts
11  +1.0 volts
• Because the NIC has to distinguish between finer
differences between signals, 1GbE is more
susceptible to noise. Many organizations opt to stick
with 100BaseT instead.
Fig. 4-15 Pulse Amplitude Modulation-5 encoding
Fiber-based Gigabit Ethernet
• 2 types of Gigabit Ethernet use fiber optic
– 1000BaseSX uses MMF and has a maximum
segment length of 220 or 550 meters depending
on the cable
– 1000BaseLX which uses MMF or SMF which
can have a segment length of up to 5 kilometers
for SMF.
• Both use 8B10B encoding, meaning data is
sent in 10 bit groups with two overhead bits.
• This data is sent at 1.25 GHz, resulting in a
net 1 Gbps data rate.
• A 10 gigabit version of Ethernet is now being
• 10GbE now has two forms, one for LANs and one
for WANs. Interconnect LAN/WAN networks will
be possible using 10GbE.
• The LAN form of 10GbE runs over four parallel
MMF or SMF fibers.
• Each fiber uses 8B10B coding at a clock speed of
3.125 GHz, yielding 3.125 x .8 x 4 = 10 Gbps.
• The WAN form operates using 64B66B (64 data
bits + 2 overhead bits) on a 10.26 GHz line,
yielding a data rate of 9.95 Gbps.
The Best Practice LAN Design
Effective Data Rates
• The effective data rate is the maximum
practical speed hardware layers can expect
to provide. It depends on 4 factors:
– nominal data rate (e.g., 10Mbps for 10BaseT)
– error rate, since this determines the frame
retransmission rate
– efficiency of the data link protocol which
depends on the percentage of the transmission
devoted to overhead
– efficiency of the media access control protocol
meaning how effective is the protocol at
making use of the nominal data rate.
Data Link Protocol Efficiency
• Based on the amount of overhead per frame and so
is the same for both shared and switched Ethernet.
• Since Ethernet frames have variable length, it also
depends on the traffic being sent over the network.
• For small frames (~150 bytes) have an efficiency
of about 82% while for large, 1,500 byte frames, it
is high, about 98%.
• Typical network traffic combines Web pages with
e-mail and usually means sending one small frame
with many large frames in the response.
• For a small packet followed by 20 large packets,
the average data link protocol efficiency would be
Media Access Control Protocol Efficiency
• Shared and switched Ethernet differ in their MAC
protocols. Shared Ethernet is very sensitive to
network traffic levels (see Figure 4-16).
– For traffic levels below 50% response time delays are
– For traffic levels between 50-80% of capacity, response
time delays are significant
– Above 80%, response time delays increase
• Thus, shared Ethernet network capacity is
effectively limited to just under 50% of capacity
• At 97% data link protocol efficiency, 10BaseT can
carry 0.97 x 0.5 x 10mbps = 4.85Mbps.
Figure 4-16 Performance of Ethernet LANs
Eff. Data Rates for Shared Ethernet
• The 4.85 Mbps estimated in the previous
slide is a shared capacity that must be
divided amongst all of the active network
• In a low traffic environment with only two
active users on a network, this would
correspond to 2.5 Mbps/user (see Fig. 4-17)
• In moderate traffic, with 10 users, this
would mean only 500 kbps/user.
• 100BaseT with 10 active users provides
about 7.5 Mbps/user.
Eff. Data Rates for Switched Ethernet
• Switched Ethernet dramatically improves
performance since each computer appears to have
its own dedicated circuit and collisions and
congestion are no longer a problem.
• Experts believe switched Ethernet users can
effective utilize 95% of network capacity.
• For a 10BaseT switched LAN, each computer
would have an effective capacity = 0.97 x 0.95 x
10Mbps ~ 9 Mbps.
• For 100BaseT switched LAN, each computer
would have an effective capacity = 0.97 x 0.95 x
100Mbps ~ 92Mbps.
Eff. Data Rates for Gigabit Ethernet
• Gigabit Ethernet is typically implemented in
a full-duplex switched environment.
• This means it provides 1 Gbps in both
directions at once.
• This makes the effective data rate for
sending data in one direction is
0.97 x 0.95 x 1 Gbps ~ 900 Mbps
• Since 1GbE is implemented as a full-duplex
network, traffic can be sent simultaneously
in both directions, so the effective capacity
is really double this value or 1.8 Gbps.
Effective Data Rate per User
Shared 10BaseT
2.5 Mbps
1 Mbps
Shared 100BaseT
37.5 Mbps
15 Mbps
7.5 Mbps
Switched 10BaseT
9 Mbps
9 Mbps
9 Mbps
Switched 100BaseT
90 Mbps
90 Mbps
90 Mbps
Full Duplex 1 GbE
1.8 Gbps
1.8 Gbps
1.8 Gbps
Full Duplex 10 GbE
18 Gbps
18 Gbps
18 Gbps
500 kbps
1. Most frames are 1500 bytes of larger
2. No transmission errors occur
3. Low traffic means 2 active users, moderate traffic
means 5 active users, high traffic means 10 active users
Figure 4-17 Effective data rates for Ethernet
LAN Recommendations
• Several best practice recommendations for LANs
are shown in Figure 4-18.
• Since network traffic almost always increase, the
best practice design is for the worst case.
– For most networks, switched 10BaseT is the best option
– For small networks, such as home networks, shared
10BaseT over cat-5 cable is best.
– For networks with high traffic levels, switched
100BaseT or 1 GbE over MMF is best.
• In most LANs, the circuit to and from the server is
the network bottleneck.
• The solution for this is to use a 10/100 switch and
then connect the server using a 100Mbps circuit.
• Most networks: Switched 10Base-T
Ethernet over category 5 cables
• Very small networks (e.g., home
networks): traditional shared Ethernet over
category 5 cables or category 3 cables
• Networks with high demands (e.g.,
multimedia networks): Switched 100Base-T
Ethernet over category 5 cables or full
duplex 1 GbE over multimode fiber
Figure 4-18: Best practice LAN recommendations
End of Chapter 4

Chapter 1. Introduction to Data Communications