Professional Documents
Culture Documents
New Doc 26-Feb-2021 12.20
New Doc 26-Feb-2021 12.20
New Doc 26-Feb-2021 12.20
D T R A NS MISSION
S AN
I
TOPOLOGIE
• I
4/
10 4 CHAPTER . l determine the nature of a LAN
.1'11�-j. .....tt'·
ology ing redients l 1 ia or M
f\N
. key t chn s:
ii·�: �b,:as follow
�. l'. .
u1� . . -i ll
• T p l gy
medi um
• Trnnsmission nt r ol tee l rn
.
1 que
ss co
• Mediulll acce . . .
an d transm ss 1 on media that ar e
. po 1ogies
s the to
1 most c
This chapter survey MANs. The issue o f access contro1 1. s bne . fly ra ised b orn
s and Ut1:s
monly use d for LAN . m Chap ter 5 an d Part Three. Th e c h apt er b e gin s With an
covere. d m· more deta. il . The n ' one sec •
tion ·
1s d evote d to p ro vid in
opo 1 og1• es. g rn
overview of the principal t 1 gies (bus/tree, rin g, and star ) . p·maIIy, the concep ore
det ail for each of these top? ?ntroduced. t of
structured cabling systems is i
In the context of a communication network, the term topology _refers to the way in
which the end systems, or stations, attached to the network a�e mterconnected. Tiie
common topologies for LANs are bus, tree, ring, and star (Figure 4.1). The bus is a
special case of the tree, with only one trunk and no branches; we shall use the term
bus when the distinction is unimportant.
J
..
Tap
--
Repeater
_flow.£Ldat!.._
- /
Terminating
.-
'
resistance
•• •
' il il
Station I /
I , , I
(a) Bus
..
(c) Ring
._ --
(\·111ral liuh, swird,,
or repeater
.,__ ___ -.
.,__ __ -..
(b) Tree
-
B C
-
A B C
A D C
Choice of Topology
ility, expand-
The choice of topology dcpends on_ a va. riety of factors, including reliab. nm . g a LAN
ability and performance. This choice is art of the over all task of destgtransm1. ss1. n
and thus cannot ·be made in isolation p dependent of the choice of �
; :r ec hnique. A few general remats ct�r:.
medium , wiring layout, and a ccess con r t d for a us
made at this point. There are fo· ur a 1 ternative media that can be use
A N deve lop . -grade tw1s
ment, v01ce . ted pair
• Twisted pair: · In the earl� da s of L of
1nexypen si ve ' ea sily installed bus AN.. A number to
plemented . s calingL. t·w1sted pair
was used to prov1'd e a n up
systems operati· ng at 1 Mbp s were im
EDIA
TO PO LO GI ES AND TRANSMISSION M
10 8 CHAPTER 4 /
■
n
(n) C trnnsmlts frnmc
addressed to A
(d) C absorbs
returning frame
yet t h'e � ar� et penet rat 10 . 11 f fib;r k k l to inst all and mai t
n ain
� of s d � er son ne
cost o f 1f b er componen t s an the lac
ille
IE AND Tl AN MIS I N MEDIA
fiber systems. Thi� situation is beginning to change rapidly as more products using fiber
are introduced.
As was mentione d, bus and tree topology LANs have been supplanted by 5 lar a:�
/
ring topology LANs. However, there is a considerable installed base 0� bus t�e
LANS, so it is worth examining this approach further. This section_ provid � s s:ith
tec hnical details on bus/tree topology LANs and MANs. The sect�o n ber n; e
an overview o f the genera l characteristics of this topology. The remainder O t
e s c
tion examines the use of coaxial cable for implementing this topology.
r
4.2 / DUS/TltEE
T P LOGY 111
r cteristics of tho Bus/Tree T opo 1 o
Cha' a gy
rf'he bus/tree topology is a multipoint config urat1.
. on That !\ . t h er�
devices connected t o ti 1_� medmm and capable of tr. are more than two
i
ansmitting on the me
to several design issues. dium . Th'J
gt·ves rise .
If more than one device at a time o n a bu
. . . . s or tree netw ork transm
ame ume, tl1 e s1gnals . w1·11 inter fere. T hus ' t 1ere .ts a requirement
i its at the
S .
. m may transmit at a to regulate which·
station on ·th ·
e medm ny po10 · t .m tim
. e: Histonc · � 11Y, t·h e most com-
.
mon multipomt access scheme has been the mu It'd I r op Im , m which access is
� dete r-
mined by polling. from a· controlli . ng station · T he c ontr oll mg stati o_n may send data
· eci. fic ah. on, askmg for an imme- /
to any other station, or 1t may issue a p oll to sp
diate response. T his metho d, ho wever negat:s o e O tt he advantages of a dis_trib-
uted system and is a wkward for comm�nication � � wee_ n two non �ontroller stati ons.
· �
A variety of distributed strategies, referred to as istnbuted medmm access control
(MAC) protocols, h ave been developed f or bus and tree topoIogies. · .
These are dis-
cussed in Part Three.
Anothe r design i_ssue has _to d o with signal balancing. When two stations
exc han�e ?ata o v�r � h?k_, the s1gnal_strength of the transmitter must be adjusted
t�. be �1th1� certam 1 �r�uts._ T�e sign�l must b� �trong enough so that, after attenu
ation across the me dium, it meets the receiver's minimum signal strength require
ments. I t must.also be stro n,g_ e_n ough to maintain an adequate signal-to-n oise ratio .
On the ot��r-�and, th� sig�al must not_ be so-strong that it overloads the circuitry
of the transmitter, which distorts the signal. Although easily done for a point-to
point link, signal balancing is ·no eas:r task for· a multip oint line. If any station
c.,an transmit to any other station, then the signal balancing must be performed for
all permutations of stati ons tak en two at a time. For n stations, that works out to
n X (n - 1) ·permutations. S o , for a· 200-station netw ork (not a particularly large
system), 39,800 signal strength constraints must be satisfied simultane ously. With
interdevice distances ranging from tens to thousands of meters, this would be an
extremely difficult task for any but small net�orks. In systems that use radio
frequency (RF) sign als, the problem is compounded because of the possibility of
de the
RF si gnal interference across f requencies. A common solution is to divi
is p ssible, using
medium into smaller segments within which pairwise balancing o
ends, where-it is absorbed (Figure 4.4a). The digi_tal signaling requires a bus topol
ogy. Unlike analog signals, digital signals cannot easily be propagated through the
branching points required for a tree topology. Baseband bus LAN systems can
•
(a) Baseband
Receive onft - - - ►
Passive
headend
◄ Transmit onft
(b) Dual cable broadband
Headend Receive onfl - - ►
frequency
converter - --- Transmit onft
◄
,. I • I
(c} Split broadband
' ,._
Figure 4.4 Baseband and Broadband Transmission Techniq
ues
-- ' , t '
4. 2 / D U S
n y a limited. distance, abou k /TREE TOPOLOGY
te d onl t1 m t 113
e�f tl1e sign w al, h i ch i s most pronoun ced a m t. This is b cc os
o pu . at h' h e f ause the att
a nd a we akening of t he en
th e l ses signa 1 to t tg r re qu en cie s causes a b uation
h e extent tha l ur
tar
ger distances
. .
is impractical · t c' 0 m municaf ring of
The ongmal use of baseband coax·ta 1 10 n over
which operates at 10 Mbps. Ethern c a b le for a b us L
te m11 1 , et b e came th e ba . AN was the Ethernet sys-
Mo st ba seband coaxial cable syste
•
s1s of the IEEE 8
m s u se a specia . 02 • 3 stan·dard
the sta n d a rd C A TV 75- 11 ca ble. These value l 50 -!1 cabl e rath er th
· s re fer to th e an.
o u g hl y sp ea k.1 ng , im pe·dance is a im
· pe d an ce of the cable.
R . · measure of how much
h ab l to ac h 1 ev a give n sign vol tage m us t be
t ec e e al strength ( A . applied to 1 '
cable su e rs le ss intense reflectio n:�; Pih:". x � ) F d1 4 . o r di _ ital signals,
the 50-.0. ff
0m
r
d
02. 3 S peci . att. ons for 10-Mbps Baseban
f1c
Table 4.3 IE EE 8
LANs
Coaxial Cable Bus
toBASES 10BASE2
10 Mbps
10 Mbps
185 m
Data rate ng t h
500m
1000m
Maximum segment le 2500m
30
Network span 100
Nodes per segme nt 2.5m 0.5m
0.5 cm
Node spacing tern
Cable diameter
114 CHAPTER 4 / TOPOLOGIES AND TRANSMISSION MEDIA
•
Segment
Repeater
.,
• • ••
Figure 4.5 Baseband Configuration
cabl e and the workstation. The cable is easier to install and uses cheaper electron
ics than the thicker cable. On the other hand, the thinner cable suffers greater atten
uation and lower noise resistance than the thicker cable. Thus it supports fewer taps
over a shorter distance.
To extend the length of the network, repeaters may be used. This device works
in a fashion somewhat different from the repeater on the ring. The bus repeateris
not used as a device a ttachment point and is capable of transmitting in both direc
tions. A repeater joins two segments of cable and passes digital signals in both direc
tions between the two segm ents. A repeater is transparen t to the rest of the syst em;
as it does n o bufferin g, it does not logically isolate on e s egment from another. So,
for exampl e, if two sta tions on different segments attempt to transmit at the same
time, their frames will in terfere with each other (collide). To avoid multipath inter
ference, only one path of segmen ts and repeaters is allowed between any two ts a·
tions. Figure 4.5 illustrates a multiple-segment baseband bus LAN.
j
4.2 / BUS/
TREE TOP
that carry the digital data. can pro pa. gat OLOGY 115
e grea ter dista
attenuation damage t he data. n ces before t
he n o.ise and
Dual and Split Configurations
As with baseband, stations on
a br adb
of a tap. Unlike baseband, however � ��d LA� attach to the cable by
me
medium; the taps that are used allow ' rlo� and is inherently a unidire i ans
. . s ig na s inserted ont ct onal
in only one duec tion. The primary reason . o th. e �ed'mm to pr
fo r t h_ is .ts that t opa gate
amplifiers that will pass signals of one fre i 1s unfeasible t o build
quency m both direct'
sional property means that only those stat tons. Th·1s urn· d1m
•
en-
. . i ons "downstream from a t,, · •
st ation can rec .
eiv e 1ts
. signals. How ' then to . ra nsmittm g
ch' .
. Clearly' two data paths are neede , ieve fu11 connectivity?
a
d These paths are Jom . .
n etwor k k no wn as the headend. For a bus · ed at a. pomt .
on the
topology, the hea'dend ts · • ply one end
sim
of the bus. For a tree topology , the headend
. . _ is the_ root of the branch'mg tree. All
stations transmit on one p�th toward the headend
(inbound). Signals arriving at the
headend are then propagated a l ong a second data path awa
y from the headend ( out
bound). All stations receive on the outbo und path.
Physically, two different configurations are used to implement
the inbound
and outbound paths (Figure s 4.4b and c). On a dual-cable configuration, the
in
bou nd and outbound paths are separate cables, with the headend simply a passive
connector between the two. Station s send and receive on the same frequency.
By contrast, on a split configuration, the inbound and outbound paths are dif
ferent frequency bands on the same cable. Bidirectional amplifiers 1 pass lower fre
quencies inbound and higher freq uencies outbound Between the inbound and
.
outbound frequency bands is a guardband, which_ carries no si�nal� and serve s
merely as a separator. The headend contains a device for convertmg mbound fre-
quencies to outbou nd freq uencies.
The frequency-conversion device at the headend can either be an analog or
d.1g1ta ·
· l dev1ce. An anal og devi·ce , known as a frequency transla tor, converts a block
· .
of frequenc1es f rom one range to another · A digital devic e, known as a remoduI a-
· .
tor, recovers the d'1g1ta · l data from the inbound analog s1gna 1 and then retransm1t· s
the data on the outbound freq uency. Thus a remodulator provides better signal
quality by re moving all of the accumuI ated nm ' •se and attenuation and transmitting
a cleaned-up signal.
. The differences b_etween sprrt d dual
configurations are minor. The split sys-
in a building. If a large
tern 1s use fu l when a single-cable :;nt is already installed
cipated then a dual-cable system
amount of bandw1'dth .ts needed : orPthe. need is anti
. . . matter of a trade-off between cost
1s md1cated. Be yond these cons1derat10n s, t•t is a
cost of the headend remodulator or
and size. The single -cable system has th f xed
makes use of more cable, taps, splitters,
frequency translator. The dual-c� bl sys; � smaller systems, where the fixed cost
and amplifiers. Thus, dual cable c_� e :r for is cheaper for larger systems, where
ble
of the headend ls noticea ble, an� sang�� ca
increme ntal costs dom inate.
. u se we have said that broadband is inherently � uni-
. ology is · con f usi n g, b eca . . . •
'Unfortunately this termm
c y broadband
1s u mdirect1ona1 • l-1 owever, th ere is no d1ff1culty
i At a g ive n {req uen tions on the cabl e.
d• rectio nal m d
e ium • nc • y bands traveling in opposite direc
in have signals in nonover1 app ·tng freque
TRANSMISSION MEDIA
116 CHAPTER 4 / TOPOLOGIES AND
Description
to two others by unidirec
A ring consists of a number of repeaters, each connecte�
lc ). Data are trans
ti nnl transmission links to form a single closed path (Figure 4.
ferred sequentially, bit by bit, around the ring from one repeater to the next. Each
repeater regenerates and retransmits each bit. ·
For a ring to operate- as a communication s network, three fun�t1ons are re
quired: data insertion, data reception, and data removal. These functions are pr0-
vided by the repeaters. Each repeater, in addition to serving as an active element on
the ring, serves as a device attachment point for data insertion. Data are transmit
ted in frames, each of which contains a destination address field; As a frame circu-
. lates past a repeater, the address field is copied to the attached station. If the station
· recognizes the address, the· remainder of the frame is copied.
Repeaters perform the data in�ertion and reception functions in a manner not
unlike that of taps, which serve as device attachment points on a bus or tree. Data
removal, however, is more difficult on a ring. For a· bus or tree, signals inserted
onto the line propagate to the end points and are absorbed by terrninato"rs. Hence,
shortly after transmission ceases, the bus or tree is clear of data. However, because
the ring ·is ·a closed loop, data will circulate indefinitely unless removed. A frame
may be removed by the addressed· repeater. Alternatively, each frame could be
removed by the transmitting repeater after it has made one trip around the loop�
The latter approach is more desirable because (1) it permits automatic acknowledg-
-
ment, and (2) it permits multicast addressing: one frame sent simultaneously to mul
tiple stations.
A variety of strategies can be used for determining how and when frames are
added to and removed from the ring. The strategy can be viewed, ·at least concep
tually, as residing in a medium access control layer, discussed in Chapter 5.
The repeater, then, can be seen to have two main purposes: (1) to contribute
to the proper functioning of the ring by passing on all the data that come its way,
and (2) to provide an access point for attached stations to send and receive data.
Corresponding-to these two purposes are two states (Figure 4.6): the listen state and
the transmit state.
I-bit delay
■
To station To From
station station
te
(b) Transmit state (c) Bypass sta
(a) Listen state
When a repeater's station has data to s'end and·when the repeater, based on
the control strategy, has permission to send, the repeater enters the transmit state.
In this state, the repeater receives bits from the station and retransmits them on its
outgoing link. During the period of transmission, bits may appear on the incoming
ring link. There are two possibilities; and they are treated differently:
1. The bits could be from the same frame that the repeater ,is still sending. This
will occur if the bit length of the ring is shorter than the frame. In this case, the
repeater .passes the bits back. to the station, which can check them as a form of
acknowledgment. . ·
2• For some cont ro l strat egies, more than one frame .could. be on- the rmg at the
a 1
·
same time. If th e repeate r., while transmitting, re. ceives bits from frame it
· d'd
. . tted later.
not ongmate, 1·t must buffer them to be transmi
. sufficient for proper ring operation.
These two sta tes, hsten a d t �ns�it are
. fu'l. In this state, a bypass relay is activated,
A thud state, the bypass state., : al;o use other than medium prop-
. - e r epeater w1' th no delay
so that sign a ls pro pag at e Past th
. en efits. (1) It provides a partial solution to
agahon. The bypass relay. affords two b. d (2)· it improves performance by e1·1m1-.
the reliability proble m, discussed 1� ter ,
not active on the network.
nating repea�er delay for those statwns ::at are
Ring Benefits
:r
. ·
. g 15 . st d access or multiaccess network (although
Like the bus and tree, the n� ����- -point links). Hence the ring shares the
the medium it.self is a collectJO� P . tobility to broadcast and incremental cost
a
same benefits as the bus/tr..,�e ' mcl udmg d by the ring ·that are not shared bY the
de
growth. There are o ther bene fit s provt'
bus/ tree topology. . int
• ortant be n e f '
it or strength of the ring is that it uses point-to-pouse
The most imp . First, beca
. . lmks. Tl 1 ere are a n urnl' er of implications of. this factcan
commumcatJOn be covered
. ch node, greater distance s
the transmitted signal is r eg ene rated at ea
11 iS CHAPTER 4 / TOPOLOGIES AND TRANSMISSION MEDIA
than with baseband bus. Broadband bus/tree can cover a similar rang
e,
amplifiers can result in loss of data integrity at high data rates. Seco but ca ade_ d
�
accommo�at� optical fiber links that provi?e_very _high data rates nd, the ri
and excellent����
tromagnellc mterference (EMI) characteristics. Finally, the electro
nics and mainte
nance of point-to-point lines are simpler than for multipoint lines.
Another benefit of the ring is that fault isolation and recovery are
for bus/tree. This is discussed in more detail later in this section. simple r tha
With the ring, th:
duplicate address problem is easily solved. If, on a bus or tree,
two stations are by
accident assig�ed the same �ddress, there _is no easy way to
sort this out. A relatively
complex algorithm must be incorporated into the LAN prot
ocol. On a ring, the first
station with an address match that is encountered by a fram
e can modify a bit in the
frame to acknowledge reception. Subsequent stations
with the same address will
easily recognize the problem.
Finally, there is the potential throughput of the ring Und
. er certain conditions,
the ring has greater throughput than a compara
ble bus or tree LAN. This topic is
explored in Chapter 15.
Timing Jitter
On a twisted pair-or coaxial cable rin LAN . .
M ' digital . 5 igna1mg
. 1�. generally used
with biphase· encoding,. typically differenti;l . anchester. As d at� circ�late around
the ring, each receiver-must recover the b'mary d ata f rom the received signal To do
. . · · · •
th1s, the receiver must know the starting and . end'mg times of each bit, so that it can
. .
sample the receive _ d signal properly. This requires that all the repeaters on the ring
be synchr?mzed, 0� cloc�ed_, together. Recall f rom Chapter 2 that biphase codes are
self-clocking; the signal includ es a transition in the middle of each bit time. Thus
each repea�er re:ove:s clocking as well as data from the received signal. This clock
r�covery will d�viate in_ a rand om fashion from the midbit transitions of the received
signal fo_r sev�ral _reasons, including noise during transmission and imperfections in
t�e rece1v�r cucmtry. The predominant reason, however, is delay distortion. Delay
_
d1st�rt10n 1s ca1:1sed by the fact that the velocity of propagation of a signal through
aries with frequency. The effect is that some of the signal com
, a gmded medium v this is known as inter
ponents of one pulse will spill over into other pulse positions;
known as timing jitter.
symbol interference. The deviation of clock recovery is
s the clockm g for two purposes: first
As each repeater receives data, itrecover
recover the data, and second, to use
to know when to sample the incoming signal to
chester signal to the next repeater.
the clocking for transmitting the differential Man
distortion. However, because the clock
The repeater issues a clean signal with no
timing error is not eliminated. Thus
ing is recovered from the incoming signal, the
tract in a random fashion as the signal
the digital pulse width will expand and con
er accumulates. The cumulative effect of
travels around the ring and the timing jitt
length, of the ring to vary. However, un less
the jitter is to cause the bit latency, or bit )
nt, bits will be drop�ed (not retransmitted as
the latency of the ring remains consta ed as the latency mcreases.
add
the latency of the ring decreases or ti on on th e number o f repeaters in a ring.
g jitter pla tes a lim ita _
Thus timin s can be
be entirely <;>ve�come, sev�ral measure
Alth ough this limitation cannot re illustrated m _Figure 4.7. First, each repeater can
taken to improve matters; these a This is a device that uses feedback to minimize
inciude a ph ase- locke d loop (PLL). ne Although the use of pbase-locked loops
the deviation fro m one bit timean to the ntary
reduces t e jitter , t 1ere s sti ll accumulation around the ring. A supplememo n-
1 th e
in one of �he repeaters, usually designated
as
measure is to inclu de a buffesr are a
written in using the recovered clock number of nd are rea d
•,itor repeater or station. Bit a cer t ai n
k. Th e buffer is initialized to hold
..
out using a crystal m-----aster cl.9.c
4.3 I RING TOP
. OLOGY
11. 9
reqmred. This .
need arises, for exa
mple, w hen a fram
. e i· s garbled bY a �rans .
lme err. or; in that.
case ' no repeater
may wis. rent
oving the ctrcul h t o assu me t h e re
rem ating frame. spons1bilitY 0 f
..
. .1ng Jitter. .
7. T1m . This is a subtle pro blem
. . hav�n
. g to d o with the clo . .
of a s1gnal m a distributed networ k . It is . k'mg
discuss ed i n th e foJJow�mg or t1mmg
subsection.
Problems 1 and 2 are reliabil ity pro blems. However, th ese two
. . problems, togeth er
with ,probl ems.
3 through 5 , can be ame 110 rate d b Ya re f'mement in the ring arch i'tec-
ture• , explained subseque .
ntly · Problem 6 1s
. a software �r o�lem, to be dealt with by
the MAC protocols discussed in Chapter 8 · problem 7 1s discussed next
.
Timing Jitter
. <?n a twisted pair or coaxial cable rin LAN ' d1.g1. tal s1.gnalmg
. •�
. generally use d
with biphase encoding, typically different'i!l M nches ter. As data circula te around
�
the ring, each receiver must recover th b ry � ta f�om the receiv� d signa l. To do
���
this, the receiver must know the starti:g I . endmg tim es o f each bit, so that it can
. .
sample the r_eceived signal properly. This requires that all the repeaters on the rin
b e s nchr?m ed, 0� cloc�ed, together. R�c�ll from Chapter 2 that biphase codes ar!
! �
self cloc�mg,. the signal mcl�des_ a tra nsi ti on in the m iddle of ea ch bit time. Thus
each repea�er re':ove�s clo�kmg as well as da ta from the received signal. This clock
��covery will deviate i� a random fashion from the midbit tra nsitions of the received
signal fo_r sev:ra� :easons, including n oise during transm ission and imperfections in
t�e rece1v�r circuitry. The predominant rea son, however, is delay distorti on. Delay
_ al through
d1st�rhon 1s caused by the fact that the velocity of propagation of a sign
e of the signal com
a guided medium varies with frequency. The effect is that som
, tions; this is known as inter
ponents of one pulse will spill over into other pulse posi
very is known as timing jitter.
symbol interference. The deviation of clock reco
e�s the clockmg for two purposes: first
As each repeater receives data, itrecov
to recover the data, and second, to use
to know when to sample the incomiqg signal
l Manchester signal to the next repeater.
the clocking for transmitting the differentia tortion. However, because the clock
no dis
The repeater issues a clean signal with
g sig nal, the timing error is not eliminated. Thus
ing is recovered from the incomin
contract in a random fashion as the signal
the digital pulse width will expand and
jitter accumulates. The cumulative effect of
travels around the ring and the timing ,
se the bit l ate ncy , or bit length, of the ring to vary. However unless
the jitter s i to c au
g re ma ins con sta nt , b its will be drop�ed (not retransmitted) as
. the latency of the rin r ses
ate y o f t e rin g de cre ase s or added as the latency mc ea . i ring.
he
t l nc h
in g jitter p l ace s a lim ita tion on the number of repeaters n a be
Thus tim e ral measures can
t be entirely �ve�com , sev�
Although this limit ation cannoare illustr ated m Figure 4 7. First, each repeater can
taken to improve matters; these L). Th is is a device that uses . fe�dback to minimize
w c ked l oo p (PL ps
inciude a p hase lo
on e bit tim e t o the ne ._Although the use of pbase-locked loory
fr
the deviation om accumulation around the ring. A suppleme nta
reduces t e jitter. t ere 1 s still aqne of he repeaters, usually designated as the mon
measure is to include a buffer in owritten �
. Bit s are in using the recovered clock and are read
,itor repeater or station tialized to hold a certain number of
out using a cry--- ----
sta l ma ster clg.ck. The buffer is ini
nth 7?:
.1.L. u CHAPTER 4 / TOPULOGJh� ANU TRANSMISSION MEDIA
Moni1or stalion
PLL 1 PLL.,_1
PLL1
bits and expands and contracts as needed. For example, the IEEE standard speci
fies a 6-bit buffer, which is initialized to hold 3 bits.That is, as bits come in, they are
placed in the buffer for 3 bit times before being retransmitted.If the received signal
at the monitor station is slightly faster than the mast�r clock, the buffer will expand,
as required, to 4, 5, or 6 bits to avoid dropping bits.If the received signal is slow, the
buffer will contract to 2, 1, or o•bits to avoid adding bits to the repeated bit stream.
Thus the cleaned-up signals that are retransmitted are p1:1rged of the timing jitter.
This combination of PLLs and � buffer significantly increases maximum feasible
ring size. The actual limit will depend on the characteris�ics of the transmission
medium, which determine the amount of delay distortion and therefore the amount
of accumulated jitter.For example, the IBM ring product specifies a maximum of
72 repeaters in a ring using unshielded twisted pair, and a maximum of 260 repe aters
in a ring using shielded twisted pair.
Star-Ring Architecture
Two observations can be made about the basic ring architecture described previ
ously. First, there is a practical limit to' the number of repeaters on a ring. As was
mentioned, a number of factors combine to limit the practical size of a ring LAN to
a few hundred repeaters.Second, the cited benefits of the ring do not depend on the
actual routing of the cables that link the repeaters.
These observations have led to the development of a refin ed ring archit. ecture
. ws the con-
the star ring , which overcomes some of the problems ·of the rmg and allo
of larg e l ocal networ ks.
.
This ar
.
chitectu re uses th e st ar wi ring strategy dis-
struction
cussed in Section 4.5. · ·
. mto a star.This is a chieved
nng .
As a first step, consi der the rearra emen t of a . e site F i.gure 4.B).This
.QYJJ,aving the interrepeater link all threads through a smge caus_e there is access to
s a number of advan a t g es . B
. ring wiring concentrator ha
I
4·3 I RING
Top OLOGY
121
Node
Node
Node
Node
• •
Nod�
Node
the signal on every ·link, it is a simple m·atter to i_solate a fault. A message can be
launched into the ring and tracked to see how far it gets without mishap. A faulty
segment can .be disconnected-no pocket full of keys needed-and repaired at a
later time. New repeaters can easily be added to the ring: Simply run two cables from
the new repeater to the site of ring wiring concentration and splice into the ring.
The bypass relay associated with each repeater can be moved into the ring
wiring concentrator. The relay can automatically bypass its repeater and two links
for any malfunction. A nice effe-cf ofthis ·reaf�re i sJllat.the__trnnsmi�sion path from
..... one working repeater to the next is approxim�t�f const��!; thus the ran-ge of signal
_
levels fo which the transmission system must automatically adapt is much smaller ..
The ring wiring concentrator greatly alleviates the ·per���ulation and instal
lation problems mentioned ea rlier. It also permits rapid recovery from a cable or
repeater failure. Nevertheless, a single failure could, at least tempor arily, disable the
entire network. Furthermore, throughput and jitter considerations still place a prac
tical upper limit on the num ber of repeaters i n a ring. Finally, in a spread-out net
work, a single wire concent .ration site dictates a lot of cable.
To attack these remaining pro blems, consider a local network consisting of
multiple rings. Each ring co_�_sis_ ts of a co�nected sequence of wiring concentrators,
and the set of rings is connected by a bndge (Figure 4.9). The bridge routes data
frames from one ring subnetwork to another., based on addressing informati on in
the frame so routed. From n physicnl point of view, each ring operates indepen.
dently of th oth r rings nttu he I to the bridge. From a logical point of view, the
bridg pr vid "trnnspnr nt routing nmong the rings.
Th bridg '.\ must l 'rf< rm five functions:
1. Input filtering. 1 -- or "l\ h ring, the bridge monitors the traffic on the ring and
opi s nil frnm \s aJ tr 'Sscd to other rings on the bridge. This function can be
p "rform d by a bridge programmed to recognize a family of addresses rather
thnn a single address.
2. Input buffering. Received frames may need to be buffered, either because the
int rring traffic is peaking or because the target output buff er is temporarily full.
3. Switching. Each frame must be routed through the bridge to its appropriate
destination ring.
....
Bridge
n ri
Node
Ring
wiring
concentnton Node
Nodt
Lobe
bypassed
Node