Chapter 14
Local Area
Networks:
Ethernet
The most common technology of the LAN
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.1 Three generations of Ethernet
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
14.1 Traditional Ethernet
» Traditional Ethernet was designed to operate on 10 Mbps.
» Accessing the network by a device through a connection method
called CSMA/CD.
» The media is shared between all stations.
MAC Sublayer
Physical Layer
Physical Layer Implementation
Bridged Ethernet
Switched Ethernet
Full-Duplex Ethernet
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.2 802.3 MAC frame
» MAC sublayer governs the operation of the access method.
» It also frames data received from upper layer and passes them to the
PLS sublayer for encoding.
» Traditional Ethernet uses 1-Persistent CSMA/CD as the access method.
» The Ethernet Frame contains 7 fields.
» Preamble, SFD, DA, SA, length/type of Protocol Data Unit (PDU),
Upper layer data and CRC.
» Ethernet does not provide any mechanism for acknowledging received
frames.
» The format of the MAC frame is shown as under:
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
802.3 MAC frame
» Preamble:
» Preamble is actually added at physical layer and is not (formally) part of
the frame.
» Contains 7 bytes (56 bits) of alternating 0s and 1s for synchronization of
the receiver.
» The pattern ONLY provides an alert and timing pulse.
»Start Frame Delimiter (SFD):
» 1 byte long (10101011) signals the beginning of the frame.
» SFD tells the receiving station that it has last chance for synchronization.
» The last bits are 11 alerting the receiver that next field is the destination
address.
» Destination Address:
» 6 bytes long and contains the physical address of the destination station
or stations.
»Source Address:
» 6 bytes long and contains the physical address of the source station
of the packet.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
802.3 MAC frame
» Length/Type:
» If length or type field’s value is less than 1518, it’s a length field and defines
the length of the data that follows.
» If the value of this field is greater than 1536, then it defines the type of the
PDU packet that is encapsulated the frame.
» Data:
» This field carries data encapsulated from the upper-layer protocols.
» It is minimum of 46 and maximum of 1500 Bytes.
» CRC:
» This field contains the error detection information.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.3 Minimum and maximum length
» Ethernet has imposed restrictions on both the minimum and maximum
lengths of a frame, as shown in the Fig 14.3 below:
» Minimum length restriction is required for the correct operation of
CSMA/CD.
» Maximum length restriction has no significance rather than historical.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.4 Ethernet addresses in hexadecimal notation
» Each station on an Ethernet Network (PC, workstations, or printer) has
its own Network Interface Card (NIC).
» The NIC fits inside the station and provides the station with a 6-byte
physical address.
» The Ethernet address is 6 Bytes long (48-bits) and is normally written in
hexadecimal notation, using a hyphen to separates bytes from each other
as shown in Fig 14.4 below:
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.5 Unicast and multicast addresses
» A source address is always unicast address.
» The destination address can be unicast, multicast or broadcast.
» A unicast destination address has only ONE recipient; the relationship
between the source to destination is one-to-one.
» A multicast destination address defines a group of addresses, the
relationship between source and receiver is one-to-many.
» The broadcast address is a special case for multicast address; the
recipients are all the stations on the network.
» A destination broadcast address is forty-eight 1s.
» Figure 14.5 below shows the destinguish a unicast address from a
multicast address.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.6 Physical layer
» Figure 14.6 below shows the physical layer for 10-Mbps Ethernet.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.7 PLS
» PLS sunlayer encodes and decodes data.
» Traditional Ethernet uses Manchester Encoding with a data rate of 10-Mbps.
» (For this data rate a bandwidth of 20 Mbaud is needed)
» Figure 14.7 below shows the functions of the PLS sublayer.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.8 AUI
» Attachment Unit Interface is a specification that defines the interface
between the PLS and MAU.
» AUI was developed to create a kind of medium independence interface
between the PLS and the MAU.
» The interface was first designed for the first implementation of Ethernet,
which used thick coaxial cable.
» The whole idea was that if in the future we want to connect the PLS
sublayer to a different MAU (using a different medium), we do not have
to change the PLS.
» Figure 14.8 below shows the AUI, AUI cable and connector.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.9 MAU (transceiver)
» Medium Attachment Unit (MAU), or transceiver , is medium-dependent.
» It creates appropriate signal for each particular medium.
» There is a MAU for each type of medium used in 10-Mbps Ethernet.
» The transceiver is a transmitter and receiver.
» It transmits signals over the medium; it receives signals over the medium.
» It also detects collisions.
» Figure 14.9 shows the position and functions of transceiver.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
MDI
» To connect the transceiver (internal / external) to the medium.
» A Medium Dependent Interface (MDI) is needed.
» The MDI is just a piece of a hardware for connecting the transceiver to the
medium.
» For external transceiver it can be tap or a tree connector.
» For internal transceiver it can be a jack.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.10 Categories of traditional Ethernet
» The standard defines four different implementations for baseband,
(digital), 10Mbps Ethernet as shown in Figure 14.10 below:
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
10Base5: Thick Ethernet
» First implementation is 10Base5, thick Ethernet, or Thicknet.
» The nick name derived from the size of the cable.
» 10Base5 was the first Ethernet Specification.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.11 Connection of a station to the medium using 10Base5
» 10Base5 uses a bus topology with external transceiver connected via tap to
a thick coaxial cable.
» Figure 14.11 below shows how a station can be connected to the medium.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.12 Connection of stations to the medium using 10Base2
» Second implementation is 10Base2, thin Ethernet, or Cheapernet.
» 10Base2 uses a bus topology with an internal transceiver or a
point-to-point connection via an external transceiver.
» Figure 14.12 below shows the connection of two system with the medium.
» (If the station uses and internal l transceiver then there is no need for an
AUI cable)
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.13 Connection of stations to the medium using 10Base-T
» Third implementation is 10Base-T, twisted-pair Ethernet.
» 10Base-T uses a physical star topology.
» The stations are connected to a hub with an internal/external transceiver.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.14 Connection of stations to the medium using 10Base-FL
» Although several types of fiber-optic 10Mbps Ethernet are defined, the
one implementation by vendors is called 10Base-FL or Fiber Link
Ethernet.
» 10Base-FL uses a star topology to connects stations to hub.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.15 Sharing bandwidth
» In an unabridged Ethernet network, the total capacity is shared between all
stations.
» If only one station wants to sends frame then it benefits the total capacity.
» But if more than one station needs to use the network then thet capacity is
shared.
» e.g. if two stations wants to send a lot of frames then probably alternate
usage of the network is done, when one sends the other refrains from
sending.
» Fig 14.15 below shows this concept.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.16 A network with and without a bridge
» The bridge divides a network into two or more networks.
» Bandwidth wise each network is independent.
» e.g. Fig 14.16 the network of 12 stations is divided into two networks with
6 stations in each.
» Now each network has the capability of 10Mbps, which is shared between
all the 7 stations (bridge is also a station) in the network segment.
» A network with heavy load, each station is theoretically offered 10/6Mbps
instead of 10/12Mbps. (assuming that traffic is not going through bridge.)
» It is obvious that if we further divides the network we’ll get more
bandwidth for each segment.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.17 Collision domains in a nonbridged and bridged network
» Another advantage of a bridge is the separation of the collision domain .
» Collision domains become much smaller and probability of the collision
is reduced tremendously with bridging.
» This concept is shown in Fig 14.17 below:
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.18 Switched Ethernet
» The idea of the bridged LAN can be extended to bridged LAN.
» Why not we have N networks, where N is the number of stations on the
LAN?
» The idea of multi port-bridge is converted to having a N-port switch.
» In this way the bandwidth is shared only between the stations and the
switch.
» In addition collision domain is also divided into N domains.
» Fig 14.18 below shows a switched LAN.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.19 Full-duplex switched Ethernet
» 10Base5 and 10Base2 has the limitation of half-duplex transmission.
(10Base-T is always a full-duplex.)
» The next step in evolution was moving from switched Ethernet to
Full-Duplex switched Ethernet.
» Full duplex increases the capacity of each domain from 10 to 20Mbps.
» Fig 14.19 below shows the full-duplex Ethernet Domain.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
No need for CSMA/CD
» In full-duplex switched Ethernet there is no need for CSMA/CD method.
» Each stations is connected to the network via two separate links.
» So each station can can send and receive independently so there is no
need for worrying about carrier sensing and collision detection.
» The job of the MAC layer is much easier.
» The carrier sensing and the collision detection functionality of MAC
sublayer can turned off.
MAC
» Traditionally Ethernet was designed as a connectionless protocol at the
MAC sublayer.
» There is no explicit flow control or error control to inform the sender that
the frame has arrived at the receiver without any error.
» When the receiver receives a frame it does not send ant positive or
negative acknowledgement.
» So to provide the flow and error control on full-duplex switched Ethernet,
a new sublayer, called MAC control is added between the LLC sublayer
and the MAC sublayer.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
14.2 Fast Ethernet
» The need for higher data rate resulted in the design of the
Fast Ethernet protocol (100Mbps).
MAC Sublayer
Physical Layer
Physical Layer Implementation
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
MAC sublayer
» The whole idea in the evolution of Ethernet from 10 to 100Mbps is to
keep the MAC sublayer untouched.
» The access method is same CSMA/CD.
» Obviously for full-duple Fast Ethernet there is no need for CSMA/CD.
» To keep the CSMA/CD backward compatibility with traditional Ethernet,
the following parameters are same for 10 and 100 Mbps Ethernets.
» Frame Format
» Minimum and maximum frame lengths
» And addressing
Auto negotiation:
» A new feature added to the Fast Ethernet allowing a station or hub a range
of capabilities.
» Auto negotiation allows two devices to negotiate the mode or data rate of
operation, it also give following capabilities:
» Allow incompatible devices to connect with one another, (connection of 10Mbps
device
with 100Mbps device )
» Allowing one device to have multiple capabilities.
» Allowing a station to check hub’s capabilities.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.20 Fast Ethernet physical layer
» Physical layer is made-up of four sub-layers.
» RS, MII, PHY and MDI.
» The reconciliation sub-layer is common to all implementations.
» The PHY and MDI are medium dependent.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Reconciliation sub-layer.
» The reconciliation sub-layer in fast Ethernet replaces the PLS sub-layer
in 10Mbps Ethernet.
» Encoding and decoding , (performed by PLS), are moved to PHY sub-
layer (transceiver), because encoding is Fast Ethernet is medium
dependent.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.21 MII
» Medium Independent Interface.
» In design of Fast Ethernet, the AUI was replaced with MII.
» MII is an improved interface that can be used with both 10 and 100 Mbps data rate.
(backward compatible with AUI).
» It features a parallel data path (4 bits at a time, forwarded by reconciliation sub-
layer) between the PHY sub-layer and the reconciliation sub-layer.
» Management functions are added in it.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
PHY (transceiver) and MDI
» The transceiver in Fast Ethernet is called PHY sub-layer.
» Besides the regular functions of transceiver in 10Mbps Ethernet it also responsible
for encoding and decoding. (function moved from PLS layer this layer.)
» A transceiver can be internal or external.
» Internal transceiver is installed closed to the medium, and connected to station via
MII cable .
» An internal transceiver is installed within the station/interface card and does not
need a MII cable.
MDI
» Medium Independent Interface.
» MDI is a piece f hardware, needed to connect the transceiver (internal or external).
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.22 Fast Ethernet implementations
» The Fast Ethernet can be categorized either a two-wire or four-wire
implementation.
» The two-wired implementation is called 100Base-X, which can be either
» Twisted-pair cable= 100Base-TX or
» Fiber-optics cable= 100Base-FX.
» Four –wired implementation is designed for only twisted-pair cable= 100Base-T4.
» So we come-up with tree implementations:
» 100Base-TX , 100Base-FX, 100Base-T4
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.23 100Base-TX implementation
» 100BAse-TX is a pairs of twisted-pair cable (either category 5 UTP or STP) in a
physical star topology.
» It allows either internal or external (with MII cable) transceiver.
Transceiver
» In Fast Ethernet Transceiver is responsible for:
» Transmitting ,
» Receiving,
» Detecting collisions and
» Encoding and Decoding of data.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.24 Encoding and decoding in 100Base-TX
» To achieve 100Mbps data rate, encoding and decoding is implemented in two
steps, as shown in Fig. 14.24 below:
» To maintain synchronization, the encoder performs block encoding.
» The four-parallel bits received from NIC are encoded to into five-serial bits using
4B/5B.
» This requires bandwidth of 125 MHz (125Mbps).
» The data rate of 125Mbps are then encoded into a signal using MLT-3.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.25 100Base-FX implementation
» 100Base-FX uses two pairs of fiber-optics cables in physical star topology.
» The implementation requires either internal or external (with MII cable) transceiver.
Transceiver
» The transceiver is responsible for the following tasks:
» Transmitting,
» Sending,
» Detecting the collision and
» Encoding and decoding.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.26 Encoding and decoding in 100Base-FX
» To achieve 100Mbps data rate, encoding and decoding is implemented in two
steps, as shown in Fig. 14.26 below:
» To maintain synchronization, the encoder performs block encoding.
» The four-parallel bits received from NIC are encoded to into five-serial bits using
4B/5B.
» This requires bandwidth of 125 MHz (125Mbps).
» The data rate of 125Mbps are then encoded into a signal using NRZ-I.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.27 100Base-T4 implementation
» (although 100BAse-TX provides data rate of 100Mbps but it requires the use of
CAT 5 UTP or STP cables, which is a limitation for the buildings which has already
been wired with voice-grade twisted-pair cable i.e. CAT-3 )
» So a new standard called 100Base-T4 was designed to use CAT-3 or higher UTP.
» 100BAse-T4 uses four pairs of UTP for transmitting 100Mbps.
Transceiver
» Transceiver function in 100BAse-T4 is similar to other implementations, but
encoding and decoding are more complex:
Encoding and Decoding
» To maintain the synchronization and at the same time reduce the bandwidth, 8B/6T
is used.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.28 Using four wires in 100Base-T4
» 8B/6T reduces the bandwidth from 100 to 75 (ratio of 8/6).
» However a voice-grade UTP is not capable of handling even this bandwidth.
» 100Base-T4 is designed to operate on 25-Mbaud bandwidths.
» For unidirectional transmission, this would requires six cable pairs. (three pair in
each direction).
» To Cut down the number of pairs to four, two-pairs are designed for unidirectional
transmission and two for bidirectional transmission.
» The two unidirectional pairs are always free in one direction to carry collision
signals.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
14.3 Gigabit Ethernet
» Recent need for an even higher data rate resulted in the design
of Gigabit-Ethernet protocol, (1000Mbps)
MAC Sublayer
Physical Layer
Physical Layer Implementation
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
MAC sub-layer
» The whole idea in evolution of Ethernet was to keep the MAC sub-layer untouched.
» However with 1-Gbps , this was no longer possible.
Access Method
» Gigabit Ethernet has two distinctive approaches for medium access.
» Half-duplex using CSMA/CD or
» Full-duplex with no need for CSMA/CD.
» Half-duplex approach is very interesting, but complicated and is not in use today.
» So almost all implementations of Gigabit Ethernet follow the full-duplex approach.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.29 Physical layer in Gigabit Ethernet
» Fig 12.29 below shows the physical layer of Gigabit Ethernet.
» The physical layer is made of 4 sub-layers.
» Reconciliation, GMII. PHY, and MDI.
» The reconciliation sub-layer is common to all implementations.
» The PHY and MDI are medium-dependent.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
RS and GMII and GMII features
RS
» The reconciliation sub-layer sends 8-bits parallel data to PHY sub-layer via a GMII
interface.
GMII
» Gigabit Medium-Independent Interface.
» It is the counter-part of MII in Fast Ethernet.
» This specification defines how the reconciliation sub-layer is connected to the PHY
sub-layer (transceiver).
» Unlike MII , GMII is not an external physical component, it does not exist outside
the NIC.
» SO in other words it is primarily logical, rather than a physical interface.
» It is a specification for integrated circuits or circuit boards for the Gigabit Ethernet
NIC.
GMII Features
» I operates at 1000Mbps, however there are chips that support both MII and GMII.
This means that station can operate on 10, 100 and 1000 Mbps using that chip.
» GMII specifies a parallel data path (8 bits at a time), between the RS sub-layer and
the transceiver.
» Management functions are included.
» There is no GMII cable and connector, as it is installed on NIC.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
PHY (transceiver) and MDI
PHY (transceiver)
» Just as in Fast Ethernet , the transceiver is medium-dependent and also encodes
and decodes.
» In gigabit Ethernet the transceiver can only be internal because there is no
external GMII to provide the connection.
MDI
» Just as in Fast Ethernet MDI connects the transceiver to the medium.
» For Gigabit Ethernet , only RJ-45 and fiber optics connectors are defined.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.30 Gigabit Ethernet implementations
» Gigabit Ethernet can be categorized as either a two-wire or a four-wire
implementation.
» The two wire implementation is called 1000Base-X which can use:
» Short-wave optical fiber = 1000Base-SX.
» Long-wave optical fiber = 1000Base-LX. Or
» Short copper jumpers = 1000Base-LX.
» The four-wire version uses twisted-pair cable (1000Base-T).
» So we come-up with four implementations as shown in Fig. 14.30 below:
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.31 1000Base-X implementation
» Both 1000Base-SX and 1000Base-LX use two fiber-optics cables.
» However 1000Base-SX uses shortwave laser and
» 1000Base-LX uses long-wave laser.
» AS Implementations are designed with an internal transceiver, so there is no
external GMAA cable or connector.
» Fig 14.31 below shows the connection of a station to the hub.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
1000Base-CX implementation
» 1000Base-CX implementation as designed to use ST cable, but it has never been
implemented.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Transceiver
» The transceiver in Gigabit Ethernet is internal.
» The functions of transceiver are:
» Encoding and decoding,
» Transmitting,
» Receiving and
» Collision detection (if appropriate).
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.32 Encoding in 1000Base-X
» To achieve 1000-Mbps data rate, encoding and decoding occurs in two steps, as
shown in Fig. 14.32 below:
1
» To maintain synchronization, the encoder performs block encoding first.
» 8 parallel bits received from NIC are encoded into 10 serial bits using 8B/10B.
» This requires a bandwidth of 1.25MHz (1.25Gbps).
2
» The data at 1.25 Mbps data rate are then encoded into a signal using NRZ
encoding.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.33 1000Base-T implementation
» 1000BAse-T was designed to use category 5 UTP.
» Four twisted pairs achieve a transmission rate of 1Gbps.
» Fig 14.33 below shows the connection of a station to the medium using
1000BAse-T.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
Figure 14.34 Encoding in 1000Base-T
» To send 1.25Gbps over four pairs of UTP, 1000Base-T uses an encoding
called 4D-PAM5 (4-dimensional, 5 level pulse amplitude modulation).
» 5 levels of pulse amplitude modulation are used.
» Fig. 14.34 below shows the overall concept.
McGraw-Hill ©The McGraw-Hill Companies, Inc., 2004
