Wired lANs: Ethemct

IEEE gTAIIDARDS

. fhe EEE Standdrd Project 802 was designed to regulate the manufacturing and interconnertivity betweendifferent LAN!,

It is a way of speclfying functions of the physical layer and the data link layer of maior l"AN protocols.

This standard was adopted by the Americon Notionol Stondords ,nstitute (ANSI).

fn 1982 the fntemational Organization for Standardization {lSO) approved the Project 802 as aninternationaf standard under the designation ISO 8802.

The relatlonshlp ofthe 8O2 Standard to the traditional OSI model is shown in Fig.

The IEEE has subdivided the data link layer into two sublayers:

a. logical link control (L[c]b. Media access control(MAc).

ll (l: L i:d Lirtcontr.l\lA(': !|dtu *e$ hrcd

t - .l ,*'"*t -l**'",t----------' tt-{icrr hyli

Iryt

i Co@ lnF.(I '

u-c

vac

Logical Link Control (llclIn IEEE Project 802, flow control, error control, and part of the framing duties are collected into

one sublayer called the logical link control.

Framing is handled in both the LLC sublayer and the MAC sublayer.

The LLC provides one single data link control protocol for all EEE LANS i.e the LLC is different fromthe media access control sublayer, which provides different protocols for different LANS. A single LLCprotocol can provide interconnectivity between different LANS because it makes the MAC sublayertransparent. Fig shows one single LLC protocol serving several MAC protocols,

Froming

LLC defines a protocol data unit (PDU). The header contains a control field used for flow and errorcontrol. The two other header fields define the upper-layer protocol at the source and destination that

2uses LLC. These fields are called the destination service access point (DSAP) and the source serviceaccess point (SSAP). The other fields defined in a typical data link control protocol such as HDLC are

-or"a tJ ttr" MaC sublayer.

In other words, a frame defined in HDLC is divided into

PDU at the LLC sublaver and

a frame at the MAC sublayer.

FiEt HoLC lrome comporcd with LLC ond MAC fromes

l)S,lP: t)c:'.i'urDn s!i.r .l..(Isr Fn"ts.SlPr Souic! s.^rrr a.a\l txrirt

t.J.c ft)t ---

llDl-C fr.l'l)c

Need for LLC

. To provide flow and error control for the upper layer protocols that actually demand services like LAN5.

. LLC may be needed to provide flow and error control for the application layer protocols.

Media Acc6s Control (MACIMedia access control defines the soecific access method for each l-AN.

For examole

o lt definea9!41qD as the med.iqlccess method for Ethernet l-ANs. Token passing method for Token Ring and Token Bus LANS.

A part ol the framing function is handled by the MAC layer. The MAC sublayer contains a number of distinctmodules; each defines the access method and the framing format specific to the corresponding LAN protocol.

Physkal Layer

The physical layer is dependent on the implementation and type of physical media used. For example,although there is only one MAC sublayer for Standard Ethernet, there is a different physical layer specification foreach Ethernet imolementations.

3SfAT{DARD ETHER ETThe original Ethernet was created in 1975 at Xerox's Palo Alto Research Center (PARCI.Four Benerations:

1) Standard Cthemet (10 Mbpsl "2) Fast Ethernet (100 Mbpsl

-

3) Glgabit Ethernet (16bps)-4) Ten-Gigabit Ethernet (10 Gbpsi *

MAC SubLyer fol Standard Eth.mct

. lt governs the operation of the access method.

. lt frames data received from the upper layer and passes them to the physlcal layer.

Fmme FormotThe Ethemet frame contains sven ftelds. lt does not provide any mechanism for acknowledging received

frames which makeJii in irnretiiUte medlum. Acknowledgments must be implemented at the higher layers. Theformat of the MAC frame ls shown in Fig.

FE:802.3 MAC frame

P'l*bk 56 hin ohtanuin! irsd fi.SI|} St'r fre (|.litnbr, i.* (l0l0l0l I )

6 byEr 2 bttes

+Preomble:

This field contains 7 bytes {56 bits) of alternatinS 0s and ls that alerts the receiving systm to the comingframe and enables it to synchronire its input timing.

Stort frome delimiter (SFD):The second field (1 bytq:.101q1011) slgnals the bgjinning-of the frame. The SFD warns the station or stations

that this is the last chqnce,for_sfnchronization. The last , bits is 11 and alens the receiver that the next field-is thedestination address.

De sti notion oddress (DA)IThe DA field is 6 bytes and contains the physical address of the destination station or stations to receive the

packet.

Source dddress (SA):The 5A field is 6 bytes and contains the physical address of the sender of the packet.

Length or typei

This field is defined as a type field or length field. The origina!thernet used this fietd as the type field todefine the upperiayer protocol using the MAC frame. The IEEE standard used it as the lengh fieiJto define thenumber oT!y1-e1i1-ihE dati tiito.

Dotd i

This field carries data encapsulated from the upper-layer protocols. lt is a minimum of 46 and a maximum of15OO bttes.

CRC I

The last field contains error detection information.

Frame Length

Ethernet has imposed restrictions on both the minimum and maximum lengths of a frame as shown in Fig.

Fig. Mlnimum ond moximum lengths

\tinitn!n. FlkE/ rmStll: a6 T-r.\Mlximu- tElbrn lserh: l5{D hylci

Dzl! rod F ding ('RC

6 bltcs 2 by,.rjM inirt:j'n franre lcngihr 5 I g b'r.i (,r 64 byts

M :rtlliflr 6?mr largar l:.1.{4 bit! o. I i l8 rryt.r

5The minimum'length restriction is required for the correct operatlon of CSMA/CD. An Ethernet frame needsto have a minimum length of 512 bits or 64 bytes. Part of this length is the header and the trailer.

tf we count 18 bytes of header and trailer (6 bytes of source addrss, 6 bytes of destination addrcss, 2 bytesof length or type, and 4 bytes of CRCI, then the minimum length of data from the upper layer is g - 18 = 45 bytes.lf the upper-layer packet is less than 46 bytes, paddint is added to make up the difference.

The standard defines the maximum length of a frame (without preamble and sFD field) as 1518 bytes. lf wesubtract the 18 bytes of header and trailer, the maximum length of the payload is 1500 bytes.

The maximum lentth restriction has two reasons,

1) Memory was very expensive when Ethernet was designed: a maximum length restriction helped toreducethe size of the buffer.

2) The maximum length restriction prevents one station from monopolizing the shared medium, blockingother stations that have data to send.

Frrme lD8th:MlnimlE:6{ b!-a (-512 bttr) Mrximurni lSlE bytcs (12,t44 birt)

NLAddresslng-.t - $'; l:

JEach station on an Ethernet network has its own networt rntetoce cord (NlC).fhe NIC provides the station

with a 6-byte physical address. As shown in Flg, the Ethernet address is 6 bytes (48 bits), normally written inhexadecimal notation, with a colon btween the bytes.

Fig: example of an Ethernet address in hexadecimal notation

O6: Ol : 02 :01 : ?( l6h,t . - l :h."d$".-an^

:4t '

Unlcast, Multicast, and Broadcast Addresss

. A source address is always a unicast address i.e., the frame comes from only one station.

o The destination address, however, can be unicast, multicast, or broadcast.

Fig shows how to distinguish a unicast address from a multicast address.

rl.,r{'ffih l'- "-*lBvh L Bl1t 2

lf the L58 ofthe first byte in a destination address is 0, the address is unicasu otherwise, it is multicast.

A unicast destination address defines only one recipient; the relationship between the sender and thereceiver is one-to-one.

6The broadcast addiess is a special case of the multicast address; the recipients are all the stations on theLAN. A broadcast destination address is forty-eight ls.

The broedcest dcsrinalion address b e spedd cesc ofthc multicgct sddress tn which all bitr are Is.

Access Method: csMA/cD

Standard Ethernet uses 1-persistent CSMAy'CD.

Slot Time :

ln an Ethernet network, the round-trip time required for a frame to travel from one end of amaximum-length network to the other plus the time needed to send the jam sequence is called the slottime.

Slot tlme = round-trip time + tlme rqulred to send the jam sequence

The slot time in Ethernet is defined in bits. lt is the time required for a station to send 512 bits. Thismeans that the actual slot time depends on the data rate; for traditional 10-Mbps Ethernet it is 51.2micro-second.

Slot Time and Collision:

512-bit slot time was chosen to allow the propr functioning of CSMA,/CD.

Consider 2 cases:

Cose 7i

Assume that the sender sends a minimum-size packet of 512 bits. Before the sender can sendthe entire packet out, the signal travels through the network and reaches the end of the network. lfthere is another signal at the end of the network (worst case), a collision occurs. The sender has theopportunity to abort the sending of the frame and to send a jam seguence to inform other stations ofthe collision.

The round-trip time plus the time required to send the jam sequence should be less than the timeneeded for the sender to send the minimum frame, 512 bits.

Cose 2 :

The sender sends a frame larger than the minimum size (between 512 and 1518 bits). In thiscase, if the station has sent out the first 512 bits and has not heard a collision, it is guaranteed thatcollision will never occur during th transmission of this frame. The reason is that the signal will reach

+the end of the nehiork in less than one-half the slot time. lf the station sends a signal on the line beforeone-half of the slot time expires then a collision occurs and the sender senses the collision. In otherwords, the sender needs to listen for a collision only during the time the first 512 bits are sent.

Slot Time and M.ximum Network Len8th

It is dependent on the propagation speed of the signal in the particular medium. ln most transmissionmedia, the signal propa8ates at 2 x 108 m/s (two-thirds ofthe rate for propagation in air). For traditionalEthernet. we calculate

Maxlength -

hopogatimftreedx gltff

Maxl-euelh = (2 x ld) x (51 .2 x 10"6 / 2i = 5120 m

lf we considerthe delay times in repeaters and interfaces, and the time required to send the jamsequence, the maximum-len$h of a traditional Ethernet network reduces to 2500 m.

6PhYslcal taver

The Standard Ethernet defines several physical layer implmentations; four ofthe most common,are shown in Fig.

ffur. &*fitl (usl.l lftltr d!i.l

Encodlng and DecodingAll standard implementations use digital signaling (baseband) at 10 Mbps. At the sender, data is

con\rerted to a digital signal using the Manchester scheme; at the receiver, the received siSnal isinterpreted as Manchester and decoded into data.

Manchester encoding is self-synchronous, providing a transition at each bit intenral.Fig shows the encodlng scheme tor Standard Ethernet,

l0 !,Cridd: r0 Mbpr d.tt

M-t s I I lld!|6c.rlf I I

I t

108ase5: Thick Ethemet (Thlckne,t I. The nlckname derives from the size ofthe cable, which is roughly the size ofa garden hose andtoo stiff to bend with your hands. lt uses a bus topology with an external transceiver connected via a tapto a thick coaxial cable.

Flg: lobases lmplementatlon

I t,,8"** lLi-l-.lo Mhli -J I - 9D tn

IBrcbdnd Cileldigital, c'|d

('dhcud'lti.l cerial crbk

rrsirnfi 5@ o

qThe transteiver is connected to the station via a transceiver cable that provides separate paths

for sending and receiving. This means that collision can only happen in the coaxial cable. The maximumlength of the coaxial cable must not exceed 500 m, otherwise, there is excessive degradation.

108ase2: Thin Ethernet ( Cheapernet )It uses a bus topology cable being much thinner and more flexible. The cable can be bent to pass

very close to the stations. In this case, the transceiver is normally part of the network interface card(NlC), which is installed inside the statlon.

Fig : 10Base2 implementation

Collision here occurs in the thin coaxial cable. lt is less expensive than thick coaxial and the teeconnections are much cheaper than taps. Installation is simpler because the thin coaxial cable is veryflexible. However, the length of each segment cannot exceed 185 m (close to 200 m) due to the hiShlevel of attenuation in thin coaxial cable.

loBase- T: Twisted-Pair Ethernet

It uses a physical star topology. The stations are connected to a hub via two pairs oftwisted cable, as shown in Fig. The two pairs of twisted cable create two paths i.e., one for sending andone for receiving between the station and the hub. Collision here happens in the hub. The maximumlength of the twisted cable here is defined as 100 m, to minimize the effect of attenuation in the twistedcable.

t0108aslF: Flber Ethernet

ft is one of the types ol opticol liber 7o-Mbps Ethernet. lt uses a star topology to connectstations to a hub. The stations are connected to the hub using two fiber-optic cables.

Fig : 7oaose-F implementotion

nm fibr{o.iccrblcs

Bascbrrd(dGat l!

loBsc-F hub

Summolv

Chamcteristics IABase5 IABasc? I0&asc-T I0Base-FMcdia Thick

coax i al cableThin

coaxial cable2 UTP 2 Fiber

Maximum length 5{X} m 185 m l00m 2fiDmLine encoding Manchcster Manchester Manchestcr Manchcstcr

C}IANGES IN THE STANDARDBrldod Ethemet

The first step in the Ethernet evolution was the division of a LAN by bridges. Eridges havetwo effects on an Ethernet LAN:

o They raise the bandwidth

. They separate collision domains

I IRaising the Bondwidth

In an unbridged Ethernet network, the total capacity (10 Mbps) is shared among allstations with a frame to send; the stations share the bandwidth of the network. lf only one station hasframes to send, it benefits from the total capacity (10 Mbps). But if more than one station needs to usethe network, the capacity is shared.

ril l -

f ' :- a l -1-l l l " '

rud i

A bridge divides the network into two or more networks. Bandwidth-wise, each network isindependent. For example, a network with 12 stations is divided into two networks, each with ostations. Now each network has a capacity of 10 Mbps. The lo-Mbps capacity in each segment is nowshared between 6 stations (actually 7 because the bridge acts as a station in each segment), not 12stations.

In a network with a heavy load, each station theoretically is offered 10/6 Mbps instead of10/12 Mbps, assuming that the traffic is not going through the bridge. lf we further divide the network,we can Sain more bandwidth for each segment. For example, if we use a four-port bridge, each station isnow offered 10/3 Mbps, which is 4 times more than an unbridged network.

Separdting Coll ision Dom ai ns

Bridge is the separates of the collision domain. Fig shows the collision domains for anunbridged and a bridged network.

i-. -il :

- i -_?-

t t )

The collision domain becomes much smaller and the probability of collision is reducedtremendously. without bridging, 12 stations contend for access to the medium; with bridging only 3stations contend for access to the medium.

Swkched Ethenet

A switched LAN has N networks, where N is the number of stations on the LAN. lt has anN-port switch. In this way, the bandwidth is shared only between the station and the switch (5 Mbpseachl. The collision domain is divided into N domains. A layer 2 switch is an N-port bridge withadditional sophistication that allows faster handling of the packets.

Fig : Switched Ethernet

Full-Duplex Ethemet

The full-duplex mode increases the capacity of each domain from 10 to 20 Mbps. Figshows a switched Ethernet in fulFduplex mode. Note that instead of using one link between the stationand the switch, the configuration uses two link: one to transmit and one to receive.

Sirirch

t3

s#rl=rlht

"#il '\.-- n --t-r

No Need for CSMA/CD

ln a full duplex switched Ethernet, each station is connected to the switch via two separatelink. Each station or swltch can send and receive indepndently without worrying about collision.

MAC Control Layer

To provide flow and error control in fullduplex switched Ethernet a new sublayer,called the MAC control is added between the LLC sublayer and the MAC sublayer.

tuFAST ETHERNET: The goals of Fast Ethernet can be summarized as follows:

Upgrade the data rate to 100 Mbps.

Make it compatible with Standard Ethernet.

Keep the same 48-bit address.

Keep the same frame format.

. Keep the same minimum and maximum frame lengths.

MAc sublayer

MAC sublayer is untouched to make it compatible with Standard Ethernet. However, a decisionwas made to drop the bus topologies and keep only the star topology. For the star topology, there aretwo choices: half duplex and full duplex.

In the half{uplex approach, the stations are connected via a hub. The access method is the (CSMA/CD)

In the full-duplex approach, the connection is made via a switch with buffers at each port and there is noneed for CSMA/CD.

Autonegotiotion

Autonegotiation is a new feature which allows a station or a hub ,a range of capabilities .lt allowstwo devices to neBotiate the mode or data rate of operation. lt was designed particularly for thefollowing purposes:

. To allow incompatible devices to connect to one another. For example, a device with a maximum capacityof 10 Mbps can communicate with a device with a lOO Mbps capacity (but can work at a lower rate).

o To allow one device to have multiple capabilities.

. To allow a station to check a hub's capabilities.

Physicdl Loyer

Topology

lf there are only two stations, they can be connected point-to-point. Three or more stationsneed to be connected in a star topology with a hub or a switch at the center, as shown in Fig

' -

-t-1

, ,2| \asf,

- -*-rbJu

I5lmplementation

Fast Ethernet implementation at the physical layer can be categorized as either two-wire orfour-wire.

Fig : Fost Ethemet implemmtotions

(4.Fr] 5 trl P Rb..

Encoding

Manchester encoding needs a 200-Mbaud bandwidth for a data rate of 100 Mbps, which makesit unsuitable for a medium such as twisted-pair cable.

Encoding for Fast Ethernet implementation are shown blow:

7U)Bose-TX uses two pairs of twisted-pair cable (either category 5 UTP or STP).For thisimplementation, the MLT-3 scheme was selected since it has good bandwidth performance but it is nota self-synchronous line coding scheme, 4B/58 block coding is used to provide bit synchronization bypreventing the occurrence of a long sequence of 0s and 1s. This creates a data rate of 125 Mbps, whichis fed into MLT-3 for encoding.

looBase-Fx uses two pairs of fiber-optic cables. Optical fiber can easily handle high bandwidthrequirements by using simple encoding schemes .lt uses the NRZ-I encoding scheme. However, NRZ]has a bit synchronization problem for long sequences of0s (or 1s, based on the encoding). To overcomethis problem, the designers used 4Bl5B block encoding as we described for 1fi)Base-TX. The blockencoding increases the bit rate from 100 to 125 Mbps, which can easily be handled by fiber-optic cable.

{q7

t'*.r ltTP r:atctd] 5

t6A looBase-Tx network can provide a data rate of 100 Mbps, but it requires the use of category 5 UTP orSTP cable. This is not cost-efficient for buildings that have already been wired for voice-grade twisted-pair (category 3).

1W)Bose-T4 uses category 3 or higher UTP. The implementation uses four pairs of UTP for transmitting100 Mbps. Encoding/decoding in 100Base-T4 is more complicated. As this mplementation uses category3 UTP, each twisted-pair cannot easily handle more than 25 Mbaud. In this design, one pair switchesbetween sending and receiving. Three pairs of UTP category 3, however, can handle only 75 Mbaud (25Mbaud) each. We need to use an encoding scheme that converts 10o Mbps to a 75 Mbaud signal. 88/57satisfies this requirement. In 88/6T, eight data elements are encoded as six signal elements. This meansthat 100 Mbps uses only (5/8) x 100 Mbps, or 75 Mbaud.

Toble: Summory of Fost EtlEnet implementotions

lO0Base-F'X IU)Bose-T4'Crt iUTPorSTP

4t.

Ca( 4 [.ITP

l(Xl mNurnbcr of wircs

t (Fm

II ury"ryShg,---_-.1_____Y!13

-L 8B/67

NTz-'

l1

clcABlT ETHERNET (1(x)0 Mbpsl Standard 802.32

The goals of the Gigabit Ethernet design are:

1. Upgrade the data rate to 1 Gbps.2. Make it comoatible with Standard or Fast Ethernet.3. Use the same 48-bit address.4. Use the same frame format.5. Keep the same minimum and maximum frame lengths.6. To support autonegotiation as defined in Fast Ethernet.

MAC Sublover

Gigabit Ethernet has two distinctive approaches for medium access i.e., half-duplex andful l duplex.

r Full-Duplex Mode

There is no collision in this mode hence CSMA"/CD is not used. Lack of collision impliesthat the maximum length of the cable is determined by the signal attenuation in the cable, notby the collision detection process.

. Half-DuDlex Mode

In this case a switch can be replaced by a hub, which acts as the common cable in which acollision might occur. The half-duplex approach uses CSMA/CD. However, the maximum length ofthenetwork in this approach is totally dependent on the minimum frame size. Three methods have beendefined: troditionol, carrier extension, ond frome bursting.

Troditionol In the traditional approach, minimum length of the frame is 512 bits.

Corrier Ertension In carrier extension approach defines the minimum length of a frame as 512bytes (4096 bits). This method forces a station to add extension bits (padding) to any frame that is lessthan 4095 bits

Frome burctino In this method Instead of adding an extension to each frame, multiple framesare sent. However, to make these multiple frames look like one frame, padding is added between theframes so that the channel is not idle.

Physlcal teyer

Topology

Gigabit Ethernet is designed to connect two or more stations. lf there are only twostations, they can be connected point-to-point. Three or more stations need to be connected in a star

l8

topology with a hub or a switch at the center. Another possible configuration is to connect several star

topologies or let a star topology be part of another as shown in Fig'

gt____-gr-t -! b. Sa.r

r. T\r! drat

d. l|t:fidty of rtrt

tmplementotion

Gigabit Ethernet can be categorized as either a two-wire or a four-wireimplementation. The two-wire implementations use fiber-optic cable (1oo0Base-sx, short-wave, or100oBase-1.x, long-wave), or sTP (10ooBase-cx). The four-wire version uses category 5 twisted-paircable (lOOOBase-T).

Fig: Gigobit Ethernet implementotions

!hol't.*!reitlr brig.r.at c fil|cr .{ppar (STP}

IqEncoding

Gigabit Ethernet cannot use the Manchester encoding scheme because it involves avery high bandwidth (2G Baud). The two-wire implementations use an NRZ scheme, but NRZ does notself-synchronize properly. To synchronize bits, particularly at this high data rate, 88/108 block encodingis used.This block encoding prevents long sequences of 0s or ls in the stream, but the resulting stream is1.25 Gbps. Note that in this implementation, one wire (fiber or STP) is used for sending and one forreceiving.

In the four-wire implementation it is not possible to have 2 wires for input and 2 foroutput, because each wire would need to carry 500 Mbps, which exceeds the capacity for cate8ory 5 TP.As a solution, 4D-PAM5 encoding is used to reduce the bandwidth. Thus, allfour wires are involved inboth input and outpuu each wire carries 250 Mbps, which is in the range for category 5 UTP cable.

Summorv

ChamcQrtttitt

Mrximurn lcng(hBlock erdrdingLinc ancoding

550m I ')ff)

m

IodJE&tt-Lx I l$fr&oso-Cx lUog.trc-TL-d 5 UTP

lmm

t(ltrg{< 5X- tatllB.c-lJc.nd l|troBd. (lX

1.25 (i!l* 1.25 Gbp.

lso I ili* d re. sfFr

Fiber Fiber$h().t.\*rec [on8-!rava

2-0

Ten-Gigabit Ethe;net (Standard 802.3ae!

The goals of the Tenigabit Ethernet design:

1. Upgrade the data rate to 10 GbPs.2. Make it compatible with Standard, Fast, and Gigabit Ethernet.3. Use the same 48-bit address.4. Use the same frame format.5. Keep the same minimum and maximum frame lengths.G. Allow the interconnection of existing LANS into a metropolitan area network (MANI or a wide areanetwork (WAN).7. Make Ethernet compatible with technologies such as Frame Relay and ATM '

MAC Subloyer

Ten-Gigabit Ethernet operates only in full duplex mode which means there is no need forcontention; CSMA/CD is not used in Ten-GiSabit Ethernet.

Physicol Loyel

The physical layer in Ten-Gigabit Ethernet is designed for using fiber-optic cable over longdistances. Three implementations are the most common: 10 GBase-S, 10 GBase-L, and locBase-E,

Table shows a summary of the Ten-Gigabit Ethernet implementaions.

CLara eristics I$GBase-L I(Xj,Base-E

I----1

Short-waveE-50-nm

rnultimode3fi) m

Long-wavel3lt}.nm

single modcl0 km

Ex.tendedl55Grnm

single mode40 km