Sie sind auf Seite 1von 170

Copyright 2003 AIRCOM International Ltd

All rights reserved


AIRCOM Training is committed to providing our customers with uality instructor led Telecommunications
Training!
This documentation is protected "y copyright! #o part o$ the contents o$ this documentation may "e
reproduced in any $orm% or "y any means% without the prior written consent o$ AIRCOM International!
&ocument #um"er' ()TR)00*)O0*+)v,
This manual prepared "y' AIRCOM International
-rosvenor .ouse
+*/0, London Road
Redhill% 1urrey R., ,L2
3#-LA#&
Telephone' 455 607 ,030 00*000
8a9' 455 607 ,030 00*000
:e"' http'))www!aircom!co!u;
UMTS Post-launch Optimisation and
Network Evolution
O056
ontents
! "ntroduction #
,!, Course Overview 0
$ %actors &imitin' apacit( !!
2!, Cell Throughput ,,
2!,!, In$luence o$ Loading $actor ,2
2!,!2 In$luence o$ 39ternal Inter$erence ,3
2!,!3 In$luence o$ 3")#o value ,+
2!,!5 The in$luence o$ orthogonality ,<
2!2 .ard =loc;ing 25
2!3 Methods o$ increasing capacity o$ a cell 2*
2!5 #etwor; Throughput 2+
2!* (rioritising Actions 20
) *educin' Mutual "nter+erence $,
3!, Introduction 2<
3!2 (ilot (ollution 30
3!2!, The .igh 1ite 30
3!2!2 Considerations when &own/Tilting 32
3!3 The Optimum >alue o$ &own/tilt 3*
- Optimisin' Network Parameters )#
5!, Introduction 30
5!2 R#C (arameters 3?
5!2!, =L3R target 3?
5!2!2 Time to trigger 5,
5!2!3 Cell reselection times 52
5!2!5 Reselection hysterisis 53
5!2!* RAC. power o$$set 55
5!3 Cell (arameters 5*
5!3!, (ilot power 5*
5!3!2 &L power per "earer 5+
5!3!3 1o$t .and Over Margin 5?
5!3!5 39ercise , *,
5!3!* 39ercise 2 *2
5!3!+ 8urther Issues regarding so$t hand over! *?
5!3!0 #oise Rise 6@L Loading 8actor7 Limit +0
5 Providin' .dditional /ard apacit( 6!
*!, Introduction +,
*!2 Channel 3lements +,
*!3 8i9ed networ; capacity +3
*!5 .igh 1peed &ownlin; (ac;et Access 6.1&(A7 ++
UMTS Post-&aunch Optimisation and Network Evolution )
."*OM "nternational &td $00)
6 "n-0uildin' Solutions 6,
+!, Introduction +<
+!2 The inter$erence loop 00
+!3 The &ead/Aone e$$ect 02
+!5 .and over $rom indoor to outdoor 05
+!5!, 3ngineering the "order region 05
+!* Implementing the in/"uilding cell 0+
+!*!, Choice o$ #ode = 00
+!*!2 &istri"ution methods ?0
+!*!3 Radiating ca"les ?,
+!*!5 8ield measurements to veri$y the implementation ?3
# Usin' Micro-cells to Service /ot Spots 1#
0!, Introduction ?0
0!2 Micro/cell and In/"uilding cells compared ??
0!3 The Theory "ehind the Micro/cell ?<
0!3!, (ilot (ower 1ettings <,
0!3!2 3ngineering the Micro/cell <2
0!3!3 :hat can go wrongB <5
0!3!5 &etecting (ro"lem Areas! <*
0!5 .otspots straddling macro cell "oundaries <+
0!* (ropagation modelling $or micro cells <0
0!+ Multiple micro cells ,00
0!0 Limiting $actors ,00
1 The E++ect o+ %urther Site Sectorisation !0,
?!, The sectored antenna ,0<
?!2 Increasing the level o$ sectorisation ,,0
?!3 @sing simulations to assess the e$$ectiveness ,,2
?!3!, >ital statistics' ,,2
?!3!2 3stimates o$ capacity ,,2
?!5 #eigh"our (lanning ,,0
, Usin' .dditional arriers 2 /ierarchical ell Structures !!,
<!, 1pectrum Allocation ,,<
<!2 &eploying e9tra carriers in the macro cell layer ,20
<!2!, A test case ,20
<!3 8i9ed networ; provisioning! ,22
<!5 Carrier loading strategy! ,23
<!* .ierarchical cell structures! ,25
<!*!, Capacity o$ micro/cells using separate carriers ,2+
<!*!2 (ilot and common channel powers in micro/cells ,20
<!*!3 Lin; "udgets $or micro/cells ,2?
<!*!5 Multi/layer strategies $or dense ur"an environments ,2<
<!*!* .and over "etween carriers ,3,
!0 "mplementin' 3iversit( S(stems !))
,0!, Introduction ,33
,0!2 &e$inition o$ 8ading ,35
,0!3 Receive &iversity ,35
,0!5 Transmit &iversity ,30
UMTS Post-&aunch Optimisation and Network Evolution -
."*OM "nternational &td $00)
,0!* Multi/@ser &etection M@& ,53
,0!+ (redicting the 3$$ect o$ &i$$erent Coverage and Capacity
3nhancement &evices ,5+
,0!0 Multiple/"eam antennas ,*2
,0!0!, =eam $orming principles ,*3
,0!0!2 Implementation in a @MT1 networ; ,*0
,0!0!3 Improvement $rom use o$ multiple/"eam antennas! ,*0
,0!? 1mart 6"eam/$orming7 antennas ,*<
!! "nte'ratin' E4tra Sites into the Macro-ell &a(er !6!
,,!, Introduction ,+,
,,!2 (lanning the new site ,+2
,,!3 Action a$ter activation o$ the new site ,+3
,,!3!, Ma;ing $urther drive/tests ,+3
,,!3!2 Assessing networ; capacity ,+5
,,!3!3 Interpreting measurements made under un;nown loading conditions!
,05
UMTS Post-&aunch Optimisation and Network Evolution 5
."*OM "nternational &td $00)
!"ntroduction
!5! ourse Overview
The objective of this two-day course is to provide delegates with
knowledge of methods, procedures and techniques that will enable
them to optimise the performance of an existing UMT !g network"
This optimisation can take the form of#
$ncreasing the network capacity
$ncreasing the provision of high-resource services
$mprove the quality of service o%ered to users
The starting point of the course is the expectation that a UMT
network has been planned, built and &optimised' to a state where it
could be launched" (overage can be described as &)*' +but
perhaps coverage for higher services may be &patchy'," -urther,
for launch, network coverage was prioritised over capacity and
therefore the capacity of the network will not be optimised"
.e can examine methods that will improve both capacity and
coverage and look at implementing them in an optimum way"
These methods will start from a &/ero resource' +simply using the
existing equipment to better e%ect, point" )ptimisation will then
involve
0e-con1guring the antenna system
2djusting cell and network parameters
3ext, an investigation into ways of utilising extra resources to
further enhance the network capacity and quality is carried out"
These resources will include#
4rovision of micro5pico-cells
-urther sectorisation
2dditional carriers
6iversity and Multi-user 6etection
2dditional macro-cell sites"
.ims o+ ourse .ims o+ ourse
C Assuming the networ; has "een
launched and'
C Coverage is DOEF!
C .igher services coverage is DpatchyF!
C Coverage prioritised over capacity
C Aim is that% at the end o$ the course you will "e
a"le to'
C Increase #etwor; Capacity
C Increase Coverage $or high resource services
C Improve uality o$ service o$$ered to users!
Introductory Session
.ims o+ ourse .ims o+ ourse
C This will include'
C DAero resourceF methods
C re/con$iguring antenna system
C adGusting cell and networ; parameters
C Adding to the networ; in$rastructure
C provision o$ micro)pico/cells
C $urther sectorisation
C use o$ e9tra carriers
C diversity and multi/user detection
C addition o$ sites in the macro/cell layer
Introductory Session
$%actors &imitin' apacit(
$5! ell Throu'hput
$f we consider the factors in7uencing the capacity of a cell, we 1nd
that this is a%ected by#
The noise bandwidth +that we cannot change,
The loading factor +that we can set within constraints,
The external interference +that we strive to minimise,
The 8b53o required on the service +that a%ects 9:80 and relies
on functionality such as power control operating well,"
The orthogonality on the downlink +that may be possible to
in7uence,
%actors &imitin' apacit( %actors &imitin' apacit(
C Cell Throughput is given "y the simpli$ied e9pressions $or pole
capacity in ;"ps multiplied "y the loading $actor
C Crucial parameters are 3")#o% inter/cell inter$erence i%
orthogonality and loading $actor 6which is a$$ected "y the
#oise Rise limit7!
Capacity Limiting Factors
( )
( )

+
i
N
E
i
N
E
b
b
1
3840
1
3840
0
0
C@plin;
C&ownlin;
$5!5! "n+luence o+ &oadin' +actor
)n the downlink, the downlink power limits the loading factor" $n
most situations, it will be possible to drive the downlink to a loading
level of about ;<=" )n the uplink it is limited in the form of a noise
rise limit" 2s the noise rise curve becomes steeper as the loading
factor gets larger, there are concerns regarding the stability of the
network if the loading level is allowed to become high" There is a
general feeling that the noise rise should be limited to about > d9
+corresponding to a loading factor of ?@=," Aowever, where the
site density is very high so that path loss is not a limiting factor, the
noise rise limit could be raised"
%actors &imitin' apacit(6 N* limit %actors &imitin' apacit(6 N* limit
C #R limit on uplin; is directly lin;ed to loading $actor'
C #R limit appears in lin; "udget and hence a$$ects coverage prediction!
C I$ a networ; is planned so that continuous coverage would "e provided
with all cells simultaneously at #R limit% then pro"a"ility suggests that
coverage is over/dimensioned!
C Coverage could "e planned $or a #R value , to 2 d= "elow the limit H
"ut this is o$ten used as a Dcom$ort $actorF margin!
C 8ailures will then "e split "etween 3")#o and #R!
Capacity Limiting Factors
10
10 1 ); 1 log( 10
NR
NR


Noise *ise vs5 Throu'hput
0!00
*!00
,0!00
,*!00
20!00
, 2 3 5 * + 0 ? < ,0 ,, ,2
Throu'hput 74!00k0ps8
N
o
i
s
e

*
i
s
e
1eries,
%actors &imitin' apacit(6 N* limit %actors &imitin' apacit(6 N* limit
Capacity Limiting Factors
1teep slope /
unsta"le
1hallow slope /
sta"le
C .owever% i$ #R is allowed to reach very high values 6e!g!I 0 d=7
there is concern that the networ; could "ecome unsta"le!
C Initially% it is e9pected that #R will "e limited to a ma9imum o$% say% +
d= until con$idence in this approach is gained!
$5!5$ "n+luence o+ E4ternal "nter+erence
8xternal interference levels are probably the most tempting target"
0educe interference and you increase capacity" Aowever, if your
main weapon +as it probably is, in reducing interference is to down
tilt the antenna, you can make hand over regions so small that
hand over failures result" This e%ect will be more noticeable if the
U8 is travelling at considerable speed" The optimum level is a
compromise" 3ote that, on the uplink, there is one value for i for
the whole cell whereas, on the downlink, each U8 experiences a
di%erent value of i.
%actors &imitin' apacit(6 %*E %actors &imitin' apacit(6 %*E
C 8reuency re/use e$$iciency is the name given to the
proportion o$ received power that comes $rom a cellJs own
users rather than $rom all users including other cells!
Capacity Limiting Factors
1
1
1
1
cell intra
cell inter
1
1
cell inter cell intra
cell intra

+

FRE
i
i
FRE
C 8reuency re/use e$$iciency is a use$ul term as it varies
"etween Aero and , as i drops $rom in$inity to Aero!
%actors &imitin' apacit(6 %*E %actors &imitin' apacit(6 %*E
C The ideal situation is where the receiving antenna can only
DseeF its own users "ut not those o$ other cells! i!e! 8R3 K ,
C The power $rom neigh"ouring mo"iles close to the cell "order
cause the "iggest pro"lems!
Capacity Limiting Factors
.igh power mo"iles close to
Cell "order cause 8R3 reduction
%actors &imitin' apacit(6 %*E %actors &imitin' apacit(6 %*E
C A large cell serving a low su"scri"er density surrounded "y
several smaller cells serving high su"scri"er densities will
e9perience a low value o$ 8R3!
Capacity Limiting Factors
A Large cell will e9perience low 8R3
=ecause it is surrounded "y
many users o$ other cells
%actors &imitin' apacit(6 %*E %actors &imitin' apacit(6 %*E
C .otspots near the cell "order will cause more pro"lems that
evenly distri"uted neigh"ouring cells
C A uantitative analysis is not always possi"le! A simulator
is e9tremely valua"le in helping to develop a $eel $or the
seriousness o$ potential pro"lems!
Capacity Limiting Factors
.ot spots near cell "order cause
8R3 reduction
%actors &imitin' apacit(6 %*E %actors &imitin' apacit(6 %*E
C Increasing 8R3' the main weapon is to down/tilt
antennas!
C Overlap o$ coverage cannot "e too small otherwise hand
over will $ail! .owever% large overlaps will lead to low
8R3!
C This is most e$$ective when there is a large angle
"etween the line $rom the antenna to the cell edge and
the horiAontal!
C In the case o$ large cells% planning to avoid hotspots near
the cell "order will reduce the incidence o$ low 8R3!
Capacity Limiting Factors
$5!5) "n+luence o+ E09No value
The 8b53o value indicates the air interface resource required by a
bearer" The higher the value of 8b53o, the lower the capacity"
8b53o is directly linked to the 9:80 and so there will be a
recommended value for a particular service" Aowever, it relies
upon the fast power control loop operating well" This is
compromised if the mobile is moving at speed through a multi-path
environment and the target 8b53o can rise by as much as < d9 in
such cases" This is something to look out for when drive testing"
The major bene1t from employing diversity techniques is the
resulting reduction in the required 8b53o over the air interface"
Multi-user detection +MU6, has a similar e%ect on the downlink"
)ptimising of network parameters, such as pilot power and soft
hand over margin, can result in a lowering of required power levels
overall"
%actors &imitin' apacit(6 E %actors &imitin' apacit(6 E
0 0
9N 9N
0 0
C .igh capacity levels depend on low levels o$ 3")#o "eing
used! 6 #ote =3R must "e accepta"le 7!
C Achieving this relies on accurate% $ast power control to
compensate $or $ast $ading!
C 8ast $ading occurs as a mo"ile moves through an
inter$erence pattern!
C Inter$erence patterns develop due to re$lections!
C Repetition distance depends on angle "etween incident
and re$lected waves!
Capacity Limiting Factors

( )

cos
2

%actors &imitin' apacit(6 E %actors &imitin' apacit(6 E
0 0
9N 9N
0 0
C This is di$$icult to estimate% $or a + d=
re$lection loss the notch depth will "e
appro9imately ,0 d=!
C 8ast power control is intended to
compensate $or the $astest $ading incidents
at the steepest slope!
Capacity Limiting Factors
E1

,
_

+
2 1
2 1
log 20
E E
E E
E2
( )

cos
2

,
_

20 /
20 /
10 1
10 1
log 20
dBdiff
dBdiff
db
Notch
%actors &imitin' apacit(6 E %actors &imitin' apacit(6 E
0 0
9N 9N
0 0
C In situations where the re$lected wave is strong% the slope o$
the standing wave pattern can "e in e9cess o$ ,00 d=)m!
C @MT1 allows $or a power up command to "e given at a rate
o$ ,*00 .A! Thus ,!* d=)ms is the ma9imum rate that can
normally "e accommodated!
C 1peeds o$ greater than ,* m)s 6*5 ;m)h7 can cause
pro"lems!
Capacity Limiting Factors
%actors &imitin' apacit(6 E %actors &imitin' apacit(6 E
0 0
9N 9N
0 0
C I$ the mo"ile cannot respond to power control commands%
the @3 will notice a variation in the received signal!
C This will lead to =3R variations that will cause the networ;
to reuire a higher target 3")#o 6a D$ast $ading marginF or
Dpower control marginF will "e reuired7!
C The e$$ect can "e to increase the target 3")#o $rom a
normal value o$ perhaps 5 d= to ,0 d= or more $or $ast
moving mo"iles!
C This will reduce the capacity o$ a cell $rom typically 32
simultaneous connections to only ? H a dramatic reduction!
C Lesson' the multipath environment and user mo"ility can
a$$ect the target 3")#o and hence cell capacity!
Capacity Limiting Factors
%actors &imitin' apacit(6 E %actors &imitin' apacit(6 E
0 0
9N 9N
0 0
C Reducing the reuired 3")#o'
C &iversity systems provide an 3")#o improvement!
C That means that the 3")#o over the air inter$ace
can "e reduced and hence the air inter$ace
capacity increases!
C Multi/user detection 6M@&7 reduces the e$$ect o$
mutual inter$erence "etween users on the uplin;!
C This reduces the reuired transmit power per user
and hence reduces the noise rise caused "y a
given num"er o$ users!
C As a result the pole capacity increases!
Capacity Limiting Factors
$5!5- The in+luence o+ ortho'onalit(
)rthogonality allows the e%ect of own-cell interference to be
reduced by the signal processing in the receiver" Typically, this sort
of interference is reduced by about > d9" The e%ect on cell
capacity depends on the value of this orthogonality factor and the
relative contribution of out of cell interference to the total level"
Most importantly, there is no guidance on positioning cells to
maximise the bene1t from orthogonality and, even if there was,
relocation of cells +which means re-doing the network plan, is not at
the top of our list"
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C &ramatic e$$ect on downlin; capacity!
Capacity Limiting Factors
( ) i +

1
N
E
3840
Capacity Pole
0
b
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C 39ample' 3")#o K 5 d=% i K 0!+% ,2200"ps
Capacity Limiting Factors
2*5?
,!0
,<,5
0!?
,*35 ,2?0 ,,00 <+3 (ole Capacity
0!+ 0!5 0!2 0 Orthogonality
(ole Capacity
6;"ps7
!000
$000
Orthogonality
055 ! 0
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C The Loading $actor delivera"le on the downlin; depends upon
the lin; loss% ma9imum transmit power and noise per$ormance
o$ the mo"ile!
C 39ample' T9 (ower 53 d=mL #oise 8loor o$ Mo"ile /,00 d=m!
C &elivera"le loading $actor can "e e9pected to e9ceed 0*M!
C (ole capacity is crucial!
Capacity Limiting Factors
( )
{ }
( )
( )
{ }
( )
( )


+

+

'



1 log 10
1 10
1
1
1 10 log 10
10
10 10
log 10 NR
d! 10 10 log 10 Po"er R# $obile
10 143
10 143
10 / 100
10 / 100 10 43
10 / 100 10 43
orth
orth LL
orth LL
orth LL
LL
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C 2uestion'
C 1uppose a group o$ users o$ a +5;"ps service in an isolated
cell e9periencing a lin; loss o$ ,3?!5 d= are demanding a total
data throughput o$ ,!025 M"ps at an 3")#o o$ 5 d=!
C :hat is the downlin; loading $actor at this throughput i$ the
orthogonality is i7 0!5 and ii7 0!?B
C 8urther% what is the tra$$ic channel power demanded and what
is the ma9imum throughput possi"le at that path loss i$ the
ma9imum tra$$ic channel power is 52!0 d=mB
C Assume a noise level at the mo"ile o$ /,02 d=m "e$ore noise
rise!
Capacity Limiting Factors
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C Answer'
C At an orthogonality o$ 0!5% the pole capacity is 2*5? ;"ps!
C ,025 ;"ps represents a loading $actor o$ 3<M!
C .ence the #oise Rise would "e appro9imately 2!23 d=!
C The e$$ective received tra$$ic power would "e /,03!0 d=m
C Actual received tra$$ic power is 2!2 d= higher 6/,0,!* d=m7
indicating a transmit power o$ 3+!?< d=m 6lin; loss ,3?!5 d=7!
C 52!0 d=m would "e a"le to deliver almost 02M loading $actor
and hence the throughput possi"le should "e appro9imately
,?32 ;"ps!
Capacity Limiting Factors
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C Answer 6continued7'
C At an orthogonality o$ 0!?% the pole capacity is 0+55 ;"ps!
C ,025 ;"ps represents a loading $actor o$ ,3M!
C .ence the #oise Rise would "e appro9imately 0!+ d=!
C The e$$ective received tra$$ic power would "e /,,0!, d=m
C Actual received tra$$ic power is 0!0 d= higher 6/,03!, d=m7
indicating a transmit power o$ 3*!3 d=m 6lin; loss ,3?!5 d=7!
C 52!0 d=m would "e a"le to deliver almost 5+M loading $actor
and hence the throughput possi"le should "e appro9imately
3*,< ;"ps!
Capacity Limiting Factors
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C Orthogonality degradation is caused "y a multipath radio
propagation environment!
C Typically% it is o$ the order o$ 0!+ in an ur"an environment% higher
in rural environments!
C In an isolated cell% an indication o$ the orthogonality can "e
o"tained "y measuring the pilot 1IR when the transmit powers o$
all channels are ;nown!
C At low values o$ path loss% all inter$erence power will "e due to
inter$erence $rom other channels!
Capacity Limiting Factors
%actors &imitin' apacit(6 Ortho'onalit( %actors &imitin' apacit(6 Ortho'onalit(
C :hat can "e done to improve orthogonalityB
C Currently% very little!
C #o guidance regarding placing o$ sites to
ma9imise orthogonality ;nown a"out!
C In $uture there may well "e "ut'/ the only outcome
would "e the recommendation to move cells 6not a
welcome recommendation as it means start
planning the networ; $rom the "eginning7!
Capacity Limiting Factors
$5$ /ard :lockin'
UMT optimisation engineers often concentrate on the air interface
or &soft' capacity of a network" Aowever, there is always a need to
ensure that there is suBcient &hard' capacity in the form of channel
elements and 1xed transmission network capacity" $t is pointless
increasing the capacity of the air interface to above the &hard'
capacity of the network" $t is tempting to launch a network with a
low level of hard capacity" $n such cases, increasing the hard
capacity of the network should be the 1rst thing to be considered as
subscriber demand grows"
%actors &imitin' apacit(6 /ard :lockin' %actors &imitin' apacit(6 /ard :lockin'
C 1o $ar we have
discussed air
inter$ace capacity or
Dso$tF capacity!
C :e could su$$er also
$rom DhardF "loc;ing
due to hardware and
$i9ed networ;
constraints!
Capacity Limiting Factors
Channel elementsB
3, lin;sB
%actors &imitin' apacit(6 /ard :lockin' %actors &imitin' apacit(6 /ard :lockin'
C There is no value in
increasing the Dso$tF
capacity o$ the air
inter$ace a"ove the
networ;Js DhardF
capacity!
C O$ten the networ; will
"e launched with a low
level o$ DhardF
capacity!
Capacity Limiting Factors
Channel elementsB
3, lin;sB
$5) Methods o+ increasin' capacit( o+ a cell
These can be divided into two categories
C" &/ero resource'#
2djusting the network con1guration in order to minimise
mutual interference
2djusting network and cell parameters in order to optimise
performance
D" &new resource'#
$nstall additional channel elements and5or increase the
capacity of the 1xed transmission network"
$mplement diversity and multi-user detection as required"
Methods o+ "ncreasin' ell apacit( Methods o+ "ncreasin' ell apacit(
C DNeroF resource'
C AdGusting con$iguration to reduce mutual inter$erence
C AdGusting networ; and cell parameters in order to optimise
per$ormance!
C #ew resource reuirements'
C Adding channel elements
C Increasing capacity o$ $i9ed networ;
C Implement diversity and)or multi/user detection!
Capacity Limiting Factors
$5- Network Throu'hput
The above equations and discussions are focused on the throughput
per cell" 3etwork capacity can be approximated as the sum of the
capacities of the individual cells" $t is therefore tempting to simply
add cells as required in order to increase capacity" 9ut, not only is
this a very expensive option, it generally leads to a reduction in the
capacities of the individual cells" Therefore the return on
investment can diminish" This is largely because the mutual
interference between cells tends to increase as cell density
increases" 3evertheless, increasing the number of cells in the
network has to be considered as the demand for capacity grows"
Aowever, it is important that the increase is managed in an eBcient
manner, implementing the solutions with maximum e%ect and
lowest cost 1rst"
Network apacit( Network apacit(
C Capacity calculations have "een Dper cellF!
C #etwor; is o$ many cells!
C Can we Gust multiply the capacity per cell "y the
num"er o$ cellsB
C &o we Gust add more cells to increase networ;
capacityB
C >ery e9pensive option
C &iminishing returns set in' higher site density results in
increasing inter$erence!
C (rocedure needs to "e structured $or ma9imum
"ene$it!
Capacity Limiting Factors
$55 Prioritisin' .ctions
2 possible order of events is#
C" 2ddress hard capacity issues
D" Use pico cells to provide in-building solutions
!" 6eploy micro cells to service hot spots
>" -urther sectorise +e"g" ? cells per site, the macro cell layer
<" 4rovision extra carriers to selected sites +in UMT a sector with D
carriers is, logically, D cells,"
?" 6eploy extra sites into the macro cell layer"
-urther, attention must be given to where cell activities 1t in with
the above list" -or example, what priority will implementing
diversity and MU6 be givenE
Network apacit( Network apacit(
C (ossi"le procedure
C Address hard capacity issues!
C @se pico/cells to provide an Din/"uilding solutionF
C &eploy micro/cells to service hot spots
C 8urther sectorise 6e!g! si9 cells per sector7
C (rovision e9tra carriers on some sites 6a sector
with 2 carriers is% logically% 2 cells in @MT17!
C &eploy e9tra sites in the macro/cell layer!
C #ote' (riority o$ deployment o$ diversity)M@&
is a topic o$ discussion!
Capacity Limiting Factors
)*educin' Mutual
"nter+erence
)5! "ntroduction
.hilst coverage may be described as very good, it is very probable
that the network capacity will be well below its limit" $t is useful to
undertake a &thought experiment' whereby you imagine a very
high user demand is present throughout your network" Fou can
then attempt to devise ways of maximising the number of
subscribers who receive a service" )f course, a Monte (arlo
simulator on a planning tool will assist in this" 9ut, generally, the
lower the mutual interference, the higher the network capacity"
Aowever, an overlap region is necessary for successful hand over to
occur and it is natural to be somewhat cautious regarding this,
initially" 3evertheless, occurrences of many pilots at a high level
+pilot pollution, must be addressed"
;*educin' Mutual ;*educin' Mutual "nte+erence "nte+erence< <
C The lower the inter$erence the higher the capacity!
C =ecause o$ the single $reuency used in a @MT1 layer% there is
an DInter$erence $eed"ac; loopF!
C This means that inter$erence% rather than Gust adding to the
"ac;ground noise level% consumes a proportion o$ the networ;
resource 6power on the &L% noise rise on the @L7!
Reducing Mutual Interference
)5$ Pilot Pollution
8ven though interference may not be high enough to produce a
noticeable e%ect on the service at network launch, the coverage
area should be investigated in order to identify areas of
interference that indicate that capacity will be limited in the near
future as demand grows" Taking action will cure the problem before
it a%ects customer service" 6rive test data can be analysed to
check for the presence of many pilots" $t must be remembered
that, in some areas, three pilots of near-equal level will inevitably
occur" More than three is in principle avoidable" $t is sensible to
start with areas where the highest number of pilots is noted" These
areas should be examined and compared with the planning tool
prediction" -rom this it should be possible to reveal which pilots are
&wanted' in that area and which are &unwanted'" The task is then
to reduce the level of the unwanted pilots so that they pose a lower
interference threat" $dentifying unwanted pilots is the easy part"
6eciding what to do about them is a harder matter" 2 few
examples are given"
)5$5! The /i'h Site
The classic text book example of a high site, where one site is much
higher than all the surrounding sites, is not very common"
Aowever, terrain features do lead to a similar e%ect" (onsider the
situation illustrated in pro1le here" The distant site produces
signi1cant interference in the service region indicated"
;/i'h Sites< ;/i'h Sites<
C O$ten% what is apparently sensi"le planning can lead to the
emergence o$ high sites!
C In the situation shown a distant site posed an inter$erence threat
in the area o$ interest!
Reducing Mutual Interference
Area o$ interest
Inter$ering
Cell
Intended
serving cell
;/i'h Sites< ;/i'h Sites<
C The $irst action to "e ta;en would "e to increase the down tilt o$
the inter$ering cell!
C Care must "e ta;en to ensure that it still provides coverage
where it is intended!
Reducing Mutual Interference
Area o$ interest
Inter$ering
Cell
Intended
serving cell
Coverage
Area
;/i'h Sites< ;/i'h Sites<
C Other possi"le solutions include reducing the cell power o$ the
inter$ering cell!
C This should "e done with great care as it will a$$ect the downlin;
coverage and capacity in its wanted coverage area!
Reducing Mutual Interference
Area o$ interest
Inter$ering
Cell
Intended
serving cell
Coverage
Area
)5$5$ onsiderations when 3own-Tiltin'
.hen down-tilting it is important to realise the e%ect on the
radiation pattern" 2ntennas will generally have a 1xed electrical
down tilt and 7exibility is in the form of mechanical down tilt" $t
should be remembered that the e%ect of mechanical down tilt
reduces as you move in a/imuth away from the principal direction"
$ndeed, at G@, mechanical down tilting has no e%ect at all" 2s the
reduction in radiation strength is surprisingly small at G@ +perhaps
CD d9,, the situation can arise where, at distance, there is a
stronger signal away from the principal direction than there is in the
principal direction" This e%ectively places a limit on the
e%ectiveness of mechanical down tilt in reducing interference"
8lectrical down tilt is, however, e%ective at reducing the radiation in
all directions" 2 typical solution to attempt would be to employ an
antenna with a 1xed electrical down tilt of perhaps ? to which can
be added a few degrees of mechanical down tilt" (are must be
taken to ensure that the cellHs wanted signal is maintained at
suBcient levels over its required coverage area and that there is
suBcient overlap to allow hand over to occur as necessary"
E4amples o+ .ntenna Tilt E4amples o+ .ntenna Tilt
/3*
/30
/2*
/20
/,*
/,0
/*
0
/3*
/30
/2*
/20
/,*
/,0
/*
0
/3*
/30
/2*
/20
/,*
/,0
/*
0
/3*
/30
/2*
/20
/,*
/,0
/*
0
No Tilt
Mechanical
3owntilt
Electrical
3owntilt
Electrical 3owntilt =
Mechanical Uptilt
Reducing Mutual Interference
&imitations on &imitations on 3owntilt 3owntilt
C I$ the antennas are
mounted centrally on a
roo$%
C The amount o$ down/tilt
achieva"le can "e limited
"y the site geometry
(lan
=loc; Image
Reducing Mutual Interference
*oo+top Main &o0e learance *oo+top Main &o0e learance
C Main lo"e is typically reuired to clear the roo$
C (arapets around roo$ edge should "e considered
C 1purious re$lections o$$ roo$ may occur i$ this is not
achieved
d K distance $rom roo$ edge 6m7
h K antenna
height a"ove roo$
6m7
p K parapet height 6m7
K tan
/,
66h/p7)d7

min
1/2 O
3d=Pvertical
4
sa$etyPmargin
4
electricalPtilt
4
mechanicalPtilt
Reducing Mutual Interference
*oo+top Plinth Mountin' *oo+top Plinth Mountin'
C (oles mounted on plinths)pads on
edge o$ roo$
C AAimuth Restrictions
C Three di$$erent ranges'
C:ithin range as $or wall
mounting o$ little concern
C@ntil aAimuth parallel to wall
Qpro"a"lyJ retain same mounting
C1hould aAimuth $ace QintoJ
structure would want to change
mounting location
C .eight 8le9i"ility
C #one with current mounting
C Reduction in height would reuire
wall mounting
C &owntilt 8le9i"ility
C #o real limit e9cept design
guidelines
C Mounting "rac;et limitations
(lan
=loc; Image
Reducing Mutual Interference
*oo+top Plinth Mountin' *oo+top Plinth Mountin'
C .eight and position o$ adGacent "uildings will also a$$ect the downtilt
possi"le
C A general rule' The principal direction o$ an antenna should
e9perience clearance o$ * metres $or a distance o$ ,00 metres $rom the
cell antenna!
Reducing Mutual Interference
)5) The Optimum >alue o+ 3own-tilt
$nterference can be said to be the limiting factor in a UMT network
and, yet, it is what makes UMT &work'# the only way to get /ero
interference is to have /ero activity" The main victim of having an
over-/ealous approach to reducing interference will be the
operation of hand over" The U8 must be able to receive at least two
good cells for the duration of the hand over procedure" 6rive tests
must be carried out to ensure the successful operation of hand over
in areas where the interference levels have been reduced by down
tilting antennas"
The optimum value will depend on likely U8 speed" The amount of
overlap required by a U8 moving at C@ kph will be much less than
for a U8 at C@@ kph" The U8 must be in the soft hand over region
for suBcient time to allow a hand over to take place" This time can
be considered to be up to a few seconds" Thus, a C@-metre region
may be suBcient for a slow-moving U8 whereas a region of C@@
metres would be required by a fast-moving U8"
The Optimum >alue o+ 3own Tilt The Optimum >alue o+ 3own Tilt
C Although a lot o$ inter$erence will reduce networ; capacity% too
little overlap can lead to hand over $ailures!
Reducing Mutual Interference
Too much inter+erence6
network capacit(
reduced5
Too little overlap6
hand over +ailures5
The Optimum >alue o+ 3own Tilt The Optimum >alue o+ 3own Tilt
C Li;ely speed o$ @3 will "e signi$icant!
C I$ 3 seconds is seen as a typical hand over time'
C @3 moving at ,0 ;ph will move appro9imately ,0 metres
C @3 moving at ,20 ;ph will move appro9imately ,20 metres!
C 1.O region must "e su$$iciently large!
Reducing Mutual Interference
Too much
inter+erence6
network
capacit(
reduced5
Too little
overlap6 hand
over +ailures5

-Optimisin' Network
Parameters
-5! "ntroduction
There are thousands of parameters that can be set, either at 03(,
3ode 9 or cell level" ome of these parameters have a more
noticeable, immediate e%ect on the performance of the network
than others" The equipment manufacturer will recommend certain
default values for initial settings" 9ut, it must be borne in mind that
the technology is very new and the recommended values are not
necessarily the optimum values" 3aturally, the initial focus is on
those parameters that have an immediate e%ect on cell
performance" These will now be considered"
Optimisin' Network Parameters Optimisin' Network Parameters
C (arameters that can "e changed'
C R#C / Dglo"alF
C #ode =
C Cell
C There are thousands o$ parameters that each
in$luence networ; $unctions!
C :e will loo; at some that are among the most
signi$icant!
Network Parameters
-5$ *N Parameters
(ertain parameters are &global' in nature, rather than being
speci1c to a certain 3ode 9, or cell" ome of these have a very
signi1cant e%ect on network performance" ome of these are
discussed below"
-5$5! :&E* tar'et
The network will strive to ensure that signal to noise ratios
delivered are suBcient to ensure that the 9:80 is low enough to
satisfy the service users" This value forms part of the outer power
control loop in which the target 8b53o value is adjusted in order to
meet the 9:80 target" 2gain, the value of this parameter is a
compromise# a very high level of 9:80 will result in dissatis1ed
service usersI a very low level of 9:80 will require a very high
target 8b53o to be imposed on the bearer with the resulting
reduction in capacity" )ptimisation can be thought of as
maximising the number of service users who assess that service as
being of &toll quality'I that is, they are willing to pay for the service"
-igures for circuit-switched traBc typically vary between @"C= and
C=" 9:80 targets for packet switched traBc can be much higher as
it is possible to allow re-transmissions of packets" The network will
be operating with greatest eBciency if the energy used in
successfully transmitting a frame of data is kept to a minimum" 2n
expression that indicates this is
FER
N E
b
1
0
where
0
N E
b
is a ratio and
FER is the frame error rate" C J FER can be thought of as the
&frame success rate'" $f this value is a minimum then the value of
8
b
per successful frame is also a minimum" 8xperiments can be
made by adjusting 8b53o and monitoring the -80" The value at
which the expression becomes a minimum depends on the
propagation conditions but is, typically, at an -80 of between D@=
and !@=" This indicates that packet traBc can be sent most
eBciently if there is a considerable re-transmission overhead but a
low signal to noise ratio" $t should be noted that the above
expression does not impose a penalty for the e%ect of delay"
:&E* Tar'et :&E* Tar'et
C 1etting a low value o$ =L3R 6e!g! 0!3M7 will produce
a high uality digital communication channel
C =ut' this will in turn reuire a high 3")#o so each
channel will use a lot o$ networ; resource!
C Optimising a networ; is supporting as many satis$ied
customers as possi"leL this means providing a
service that is DGust good enoughF!
C Increasing the =L3R target 6to% say% ,M7 will
increase capacity!
C Assessment on impact is o$ten su"Gective!
Network Parameters
:&E* Tar'et - Packet Services :&E* Tar'et - Packet Services
C >alues o$ 0!3M and ,M are typical $or Circuit
1witched 6C17 services 6e!g! voice% >T7!
C 8or pac;et switched 6(17 services% delays and re/
transmissions can "e tolerated!
C =L3R can "e much higher!
C Crucial parameter'
C This indicates the energy reuired to success$ully
trans$er a $rame o$ data!
C #etwor; is DoptimisedF i$ this is a minimum!
Network Parameters
FER
N
E
b
1
0
:&E* Tar'et - Packet Services :&E* Tar'et - Packet Services
C >alue o$ 83R $or minimum depends on radio channel!
C Typically 20M /I 30M!
Network Parameters
FER
N
E
b
1
0
FER
$0? )0?
*adio han !
*adio han $
:&E* Tar'et - Packet Services :&E* Tar'et - Packet Services
C 8or pac;et services% it appears that low 3")#o% high 83R
is the most e$$icient situation!
C Re/transmission strategy)protocol reuired!
C Method does not impose a penalty $or delay' assumes
that in$ormation is per$ectly Ddelay tolerantF!
Network Parameters
-5$5$ Time to tri''er
$n order to avoid an excessive amount of network management
activity, the network does not react instantaneously to
measurements indicating that event thresholds have been crossed"
There is a built in &time to trigger' +TTT, that is a delay time to
ensure that anomalous readings do not cause unwanted updating of
the active set, for example" 2gain, the time is a compromise# too
short and the network will experience very frequent attempts to
enter or leave soft hand overI too long and soft hand over may be
delayed by long enough to cause interference problems" )ne
signi1cant question related to this is# &Aow far will the U8 move
during the triggering timeE'"""$f the U8 moves a long way into
another cellHs coverage area before that cell becomes a member of
the active set, then interference will result" oft hand over exists to
allow the new cell to control the U8 power levels" Typical values of
TTT are from C@@ ms to C@@@ ms"
Time to Tri''er 7TTT8 Time to Tri''er 7TTT8
C #etwor; $unctions 6such as active set updates7 are
triggered "y DeventsF!
C 8or e9ample' 3vent ,C'/ DA non/active (rimary C(IC.
"ecomes "etter than an active (rimary C(IC.F!
C This event would normally trigger an active set update!
C .owever% it is important incorporate a TTT "e$ore the
update is e9ecuted!
Network Parameters
Time to Tri''er 7TTT8 Time to Tri''er 7TTT8
Network Parameters
CI$ TTT is too
small% too many
hand over
reuests will "e
made 6.O is a
Dris;y timeF7
time
R1C(
Active pilot
#on/active pilot
CI$ TTT is too
long% non/active
cell will cause%
and su$$er $rom%
inter$erence!
C Typical TTTs' ,00 ms to ,000 ms
-5$5) ell reselection times
.hen in idle mode, the mobile keeps track of its &best server'"
)nce it has detected that a new cell is a better server than its
existing cell, a re-selection procedure will be instigated" There is,
however, a reselection delay before this happens" This delay is
typically a few seconds" $f it is too small, there will be frequent
reselections" $f it is too long, the U8 may be in an inappropriate
location when it makes a call attempt" This will lead to higher than
necessary interference"
ell *e-selection 3ela( ell *e-selection 3ela(
Network Parameters
C 1imilar to TTT% "ut re$ers to idle mode operation!
C Only one cell is Dcamped onF at a time!
C I$ delay is too short% $reuent re/selections will occur!
C I$ delay is too long% @3 will attempt to ma;e a connection
on a lin; with high path loss! This will result in inter$erence!
C @sually longer than TTT / typically two or three seconds!
-5$5- *eselection h(sterisis
$n order to avoid a &ping pong' e%ect in idle mode, a hysterisis
margin is used to force the U8 to tend of stay with its serving cell"
This is similar to the approach used in KM networks" The value of
the margin is typically a few d9" $f it is too small, there will be an
excessive amount of reselection attemptsI if it is too large, there is
a danger that the U8 will 1nd itself in a poor quality radio channel
when it attempts to make5receive a call"
ell *e-selection /(sterisis ell *e-selection /(sterisis
Network Parameters
C &esira"le to have cell camping onto D"est serverF "ut'
C need to avoid Dping pongF e$$ect!
C .ysterisis is inserted to ma;e @3 tend to Dhang onF to
selected cell 6similar concept to that employed in -1M7!
C 3$$ect o$ getting it wrong'
C too small'/ Dping pongF re/selections
C too large'/ @3 can "e in poor radio environment when it attempts a
call!
-5$55 *./ power o++set
.hen establishing a call the U8 sends a 02(A request on the uplink
+U:," The power level of the 02(A request is determined by
assessing the power of the pilot 0(4 on the downlink +6:," $t is
possible to set a parameter that modi1es the U: transmit power"
2gain an optimum value needs to be found" $f the 02(A power is
too high, the uplink traBc channels will su%er from interference and
capacity will be reducedI if the power is too low then call set up
may fail" The o%set is typically from /ero to J< d9"
*./ Power O++set *./ Power O++set
Network Parameters
C Call set up is a vulnera"le time!
C 8ast power control not esta"lished!
C @3 DguessesF appropriate power level $or initial RAC.
C Too low / not heard "y cell% set up ta;es a long time or $ails
C Too high / @L inter$erence results
C RAC. power o$$set adGusts initial guess!
C@3 measures C(IC. R1C(
C3stimates path loss
CRAC. power level
determined using
estimate o$ path loss
-5) ell Parameters
(ertain parameters need to be optimised for the particular cell and
its environment" ome will depend on quite straightforward issues,
such as the feeder loss" The optimum value of other parameters
will depend on less de1nable inputs such as average U8 speed"
ome examples are given"
-5)5! Pilot power
The pilot power is set at the &rack' output" Aowever, if masthead
ampli1ers +MA2s, are used, then soft hand over gain is maximised
if the pilot strength at the masthead is equal" Therefore, it is
common in such circumstances to, initially, adjust output powers so
as to make them equal at the masthead" :ater in the optimisation
process, it is common to allow di%erent cells to accept di%erent
levels of loading" Tailoring of the pilot powers is necessary to
ensure that this is e%ected without problem"
ell Pilot Power ell Pilot Power
Network Parameters
C (ilot power dictates'
C cell coverage
C so$t hand over regions
C @L so$t hand over gain is ma9imised i$ @L path loss is
eual!
C I$ M.As are employed% pilot power should "e eual at the
mast head% not at (A output!
-5)5$ 3& power per 0earer
2s well as allocating a maximum total power to traBc channels, it is
possible to specify a limit to the power that can be allocated to any
one bearer" The most appropriate value of this parameter is
dependent on cell activity" $f the cell is very quiet then a large
allocation will reduce the probability of the downlink failing" 2s the
cell is quiet, capacity issues will not arise" $f, however, there is a lot
of demand for services, lowering the maximum allocation will
ensure that those with the lower demand +i"e" U8s in areas of low
interference, will get priority" This will lead to a maximisation of
throughput, and hence revenue" $t is possible to specify a di%erent
maximum level for active bearers compared with the level for call
set up"
Ma4imum 3& Power per :earer Ma4imum 3& Power per :earer
Network Parameters
C &L users share the power availa"le to tra$$ic channels
6typically ,+ :atts or 52 d=m7!
C Allowing one user to use all this power would mean the cell
is "loc;ed to other usersL a limit is imposed!
Ma4imum 3& Power per :earer Ma4imum 3& Power per :earer
Network Parameters
C I$ demand is low% it is "est to have a high limit so that
coverage is ma9imised!
Ma4imum 3& Power per :earer Ma4imum 3& Power per :earer
Network Parameters
C I$ demand is high% a lower limit will ma9imise throughput
6and hence revenue7 at the e9pense o$ coverage in areas
o$ high path loss or inter$erence!
-5)5) So+t /and Over Mar'in
-5)5)5! Macro-diversit( @ Ma4imal om0inin' Aain
oft Aand over is a necessity in any single-frequency cellular
network" $n a multi-frequency technology, such as KM, the
possibility exists to ensure that the &new' connection has a
signi1cantly lower path loss than the &old' connection before hand
over takes place" $n a single frequency network, the resulting
interference on the &new' cell would drastically reduce the capacity
of the network" oft hand over entails the mobile simultaneously
connecting with more than one cell" 2lthough the most signi1cant
purpose of introducing oft Aand over was to reduce uplink
interference, there are other bene1cial e%ects" -irstly, when more
than one path is provided for the radio link, a diversity gain is
obtained" There is a low probability of both channels su%ering a
bad fade simultaneously" Thus there is a reduced need for a margin
to accommodate such fades" $n this way, the target 8
b
53
@
value can
be reduced when in soft hand over" This is true of both the uplink
and the downlink"
$n addition to the diversity +or &macro-diversity', gain a%orded, the
receiver in the mobile +and the receiver at a 3ode 9 that is used
when two cells from the same 3ode 9 are in soft, or rather &softer'
hand over, processes the multiple received signal to produce and
output that is of higher quality than any individual signal" The
result on the uplink is that the transmit power of the mobile can be
substantially reduced when in soft hand over J having bene1cial
e%ects for coverage and interference" )n the downlink, providing
additional hand over channels places a power burden on the cell"
This is partially +but usually not fully, o%set by reduction in the
target 8
b
53
@
value" The general conclusion is the oft Aand over
assists the uplink but places an additional burden on the downlink"
The amount of use made of soft hand over a%ects the relative
capacities of the two directions"
So+t /andover So+t /andover
C As well as providing vital power control $unctionality% 1o$t .andover
improves the uality o$ the channel "y means o$ two methods!
C Macro/diversity -ain
C Ma9imal Com"ining -ain
Soft Handover
Macro-3iversit( Aain Macro-3iversit( Aain
C I$ the mo"ile communicates with
more than one cell% protection
against $ailure is provided as this
$ailure would have to occur on all
lin;s to cause a call to drop!
C As the "etter uality lin; can "e
selected% there is less variation in
overall channel uality!
C This leads to a reduction in (ower
Rise H the increase in average
transmit power that occurs as a
mo"ile responds to power control
commands!
-5
0
5
!0
!5
$0
$5
Mo0ile T4 Pwr .vera'e Non-+adin'
Power *ise
Soft Handover
Macro-3iversit( Aain Macro-3iversit( Aain
C The reduction in (ower Rise
helps to increase uplin;
capacity as the average T9
power is reduced!
-5
0
5
!0
!5
$0
$5
Mo0ile T4 Pwr .vera'e Non-+adin'
*educed Power *ise +ollowin' Macro 3iversit( Aain
Soft Handover
So+t /andover 2 om0inin' the Si'nals So+t /andover 2 om0inin' the Si'nals
C On the @plin; there are two possi"le methods o$ com"ining the two
6or more7 signals!
C :hen the two cells are on separate sites 6conventional Dso$tF
handover7% the R#C simply selects the "etter o$ the two signals!
C :hen the two cells are on the same site 6Dso$terF handover7%
ma9imal com"ining o$ the two signals can "e implemented!
C Ma9imal com"ining leads to an output that is o$ "etter uality 6 less
noisy 7 than either o$ the individual signals!
C Ma9imal com"ining is implemented in the mo"ile to com"ine the
downlin; signals!
C Macro/diversity gain and Ma9imal com"ining gain com"ine to
produce 1o$t .andover -ain!
Soft Handover
So+t /andover 2 Ma4imal om0inin' So+t /andover 2 Ma4imal om0inin'
C Consider the case where two signals arrive at the inputs to a
com"iner! One is DgoodF 6e!g! 3")#o K ? d=7 and the other is DpoorF
63")#o , d=7!
C It is possi"le to com"ine the signals such that the output has an
3")#o greater than ? d=! This reuires correct 6Dma9imalF7 weighting
o$ the two signals!
C3")#o ?
d=
C3")#o ,
d=
CBB
Soft Handover
So+t /andover 2 Ma4imal om0inin' So+t /andover 2 Ma4imal om0inin'
C The 3")#o at the output when the inputs are ma9imally
com"ined is given "y the simple $ormula!
C It must "e noted that 3")#o is uoted as a ratio 6not in d=7!
C ? d= corresponds to +!3 as a ratio!
C , d= is a ratio o$ ,!2+!
C These sum to 0!*+ which is ?!? d=!
2
0
1
0 0

,
_

,
_

,
_

N
E
N
E
N
E
b b
out
b
Soft Handover
-5)5- E4ercise !
.hat 8b53o improvement is o%ered when two signals of equal
quality are combined E
2nswer #-
-5)55 E4ercise $
.hat is the 8b53o at the output of a combiner if the input is
composed of two signals one with an 8b53o of ? d9 and the other
with and 8b53o of -D d9E
2nswer#
Ma4imal om0inin' - Buestions Ma4imal om0inin' - Buestions
2uestion ,'
C :hat 3")#o improvement is o$$ered when two signals o$
eual uality are com"inedB
Soft Handover
Ma4imal om0inin' - solution Ma4imal om0inin' - solution
Answer ,'
C As the 3")#o o$ the two inputs are eual% the 3")#o at the
output will "e Ddou"leF that at the input!
C This is an improvement o$ 3 d=
2uestion 2'
C :hat is the 3")#o at the output o$ a com"iner i$ the input
is composed o$ two signals' one with 3")#o o$ + d= and
the other with an 3")#o o$ /2 d=!
Soft Handover
Ma4imal om0inin' - solution Ma4imal om0inin' - solution
Answer 2'
C + d= is a ratio o$ 3!<?! /2 d= is a ratio o$ 0!+3!
C These sum to 5!+,!
C This is eual to an 3")#o o$ +!+5 d=
Soft Handover
)ptimising oft Aand )ver 4arameters
Optimisin' So+t /andover Parameters Optimisin' So+t /andover Parameters
C The parameter o$ most signi$icance is the 1o$t .andover
DAddF and DRemoveF :indows!
C They in$luence the num"er o$ terminals in so$t handover!
C -enerally% the larger the window is made% the lower the
loading on the uplin; and the higher the loading on the
downlin;!
C The path loss at the cell edge will in$luence the optimum
value o$ the 1.O window!
C The lower the path loss the larger the value can "e 6as the
downlin; will pro"a"ly have plenty o$ spare power availa"le7!
Soft Handover
Optimisin' So+t /andover Parameters Optimisin' So+t /andover Parameters
C The amount o$ improvement on the uplin; and loading on the
downlin; depends on the amount o$ so$t handover gain
achieved!
2 d=
window
5 d=
window
Soft Handover
Optimisin' So+t /andover Parameters Optimisin' So+t /andover Parameters
C 1uppose each terminal shown a"ove represents a +5 ;"ps
5 d= 3")#o connection!
2 d=
window
5 d=
window
Soft Handover
Estimatin' %*E Estimatin' %*E
C 1uppose the terminals are arranged in groups o$
5 with the path loss to the two #ode =s changing
in , d= increments!
C The red terminals will each cause an inter$erence
level , d= less than the wanted signals'
euivalent to the load o$ 3 terminals!
C The orange terminals will each cause an
inter$erence level 3 d= less than the wanted
signals' euivalent to 2 terminals!
C 6 ,d= less to their #ode =% and 2 d= less to ours 7
C Total inter$erence load' * euivalent terminals!
8R3 K +2!*M 6*)?7
2 d=
window
5 d=
window
Soft Handover
Estimatin' %*E and &oadin' Estimatin' %*E and &oadin'
C 3")#o is 5 d=
C (ole Capacity K <<* ;"ps
C Loading K *5M 6#RK3!5 d=7
2 d=
window
5 d=
window
Soft Handover
Estimatin' The E++ect o+ S/O Estimatin' The E++ect o+ S/O
C Assumptions'
C :indow set to 5 d=!
C 1.O allows the @L T9 power to reduce "y ,!* d=
63$$ectively ma;ing the 3")#o 2!* d=7!
C 1.O allows the target 3")#o on the &L to "e reduced!
This is assumed to "e 2 d= 6ma9imal com"ining on
downlin;7!
C =@T downlin;s must service twice the num"er o$
terminals 6a 3 d= e9tra "urden7!
C 1ummarising the e$$ect' @L loading $actor will reduce
$rom *5M to 3?M! #R will reduce $rom 3!5 d= to 2!,
d=! &ownlin; T9 (ower will increase "y appro9imately
, d=!
2 d=
window
5 d=
window
Soft Handover
Estimatin' The E++ect o+ S/O Estimatin' The E++ect o+ S/O
C I$ the window is set to 2 d=!
C The &L will only have to su$$er an increase o$ *0M in
the num"er o$ terminals 6to ,27 and ? o$ these will
"ene$it $rom 1.O gain! Overall increase in "urden
estimated to "e 0!* d=!
C @L split "etween users with a target 3")#o o$ 2!* d=
and those with 5 d=! Com"ined loading estimated to
"e 20M 4 ,<M K 5+M
C 1ummarising the e$$ect' @L #R will reduce $rom
3!5 d= to 2!0 d=!
C &ownlin; T9 (ower will increase "y 0!* d=!
2 d=
window
5 d=
window
Soft Handover
Estimatin' The E++ect o+ S/O6 onclusion Estimatin' The E++ect o+ S/O6 onclusion
C 1etting the window to the optimum siAe can
"alance the uplin; and downlin; in a networ;!
C #ote that e9ample here is with symmetrical
loading! 39cessive 1.O reduces the a"ility $or
the &L to serve asymmetric users!
C #ote also that 1.O reuires additional
hardware in the #ode = to provide the
necessary "earers!
2 d=
window
5 d=
window
Soft Handover
-5)56 %urther "ssues re'ardin' so+t hand over5
2lthough soft hand over produces a macro-diversity and a
processing gain, in order to be successful the receiver must
synchronise using the incoming pilot signal" $f a U8 is to bene1t , all
active set pilots must be detectable" 6etectable is usually taken to
mean an 8c5$o better than -C< d9" $f we put ourselves in the
position of a U8 that is attempting to synchronise with a cell that is
just strong enough to be in the active set, the results are
interesting" uppose that the primary server pilot is detected with
an 8c5$o of -CD d9 at the cell edge" (learly a large soft hand over
window will be of little use" This would lead to attempts to admit
cells with such a weak pilot that no use could be made of the
signals" Unless there are exceptional circumstances, soft hand
over windows5margins should be no larger than ? d9"
S/O6 +urther issues S/O6 +urther issues
C 3ven at the edge o$ the 1.O region! All cell pilots must "e o$ a
Duse$ulF strength 6typically% I/,* d=7
C Large 1.O windows not e$$ective as wea;er pilot is not detecta"le!
C In e9ceptional circumstances% larger 1.O windows may "e
e$$ective!
2 d=
window
5 d=
window
Soft Handover
S/O6 +urther issues S/O6 +urther issues 2 2 pilot power pilot power
C :e have seen that 1.O provides diversity and com"ining gain%
particularly on the uplin;! This gain helps with uplin; coverage!
C The ma9imum "ene$it is o"tained i$ the two signals are at near/
eual strength! Thus the hand over region should "e where the
uplin; path loss is the same to cells in the active set!
Ma9imum @L
1.O gain
where @L path
loss is eual
Soft Handover
S/O6 +urther issues S/O6 +urther issues 2 2 pilot power pilot power
C .and over region is controlled "y pilot power!
C I$ pilot powers are not eual% so$t hand over region will not "e
where ma9imum gain is provided!
Ma9imum @L
1.O gain
where @L path
loss is eual
Soft Handover
(ilot power controls
location o$ 1.O
region
S/O6 +urther issues S/O6 +urther issues 2 2 pilot power6 pilot power6 M/.s M/.s
C I$ M.As are deployed the @L Dends at the mast headF!
C (ilot power should "e eual at the mast head!
C I$ no M.As% pilot power should "e eual at TR9 output 6rac;
output7!
Ma9imum @L
1.O gain
where @L path
loss is eual
Soft Handover
(ilot power controls
location o$ 1.O
region
-5)5# Noise *ise 7U& &oadin' %actor8 &imit
)n the downlink, the downlink power limits the loading factor" $n
most situations, it will be possible to drive the downlink to a loading
level of about ;<=" )n the uplink it is limited in the form of a noise
rise limit" 2s the noise rise curve becomes steeper as the loading
factor gets larger, there are concerns regarding the stability of the
network if the loading level is allowed to become high" There is a
general feeling that the noise rise should be limited to about > d9
+corresponding to a loading factor of ?@=," Aowever, where the
site density is very high so that path loss is not a limiting factor, the
noise rise limit could be raised"
5Providin' .dditional /ard
apacit(
55! "ntroduction
9ecause the air interface of a UMT network is, perhaps, the most
challenging area to optimise, the need to ensure that there is
adequate &hard' capacity to support the &soft' capacity of the air
interface is sometimes overlooked" Aowever, the way in which
bearer form require channel elements and the loading cell users put
onto the 1xed network must be considered" -urther, the
modulation scheme used in .-(6M2 imposes a hard limit on the
maximum throughput" 3ew schemes are being put forward that
will increase possible throughputs"
55$ hannel Elements
8very radio bearer requires an associated bit of hardware within the
3ode 9" This hardware is referred to as a channel element +(8,"
The di%erent types of radio bearer serviced by the 3ode 9 will
require a di%erent level of resource" Typically, the requirement
could be#
Loice# C channel element per connection
?> kbps data# ! channel elements per connection
CD; kbps data# < channel elements per connection
!;> kbps data# ; channel elements per connection
The allocation of (8s is to a 3ode 9" This allocation would be
shared between cells and between carriers +if more than one carrier
is deployed," -urther, it should be noted that some of the capacity
will be required to service soft hand over connections +but not
softer hand over,"
;/ard< apacit( ;/ard< apacit(
Hard Capacity
C =ecause the air inter$ace in @MT1 networ;s is new% most
attention is paid to ma9imising the inter$erence/limited
capacity o$ the air inter$ace itsel$ 6the Dso$tF capacity7!
C .owever% there must "e su$$icient capacity in the hardware
o$ the $i9ed networ; to support the demand generated "y
the cells!
C This includes considering'
C Channel 3lements at the #ode =
C Capacity o$ the inter$aces
C Capacity o$ the R#Cs
;/ard< apacit( - hannel Elements ;/ard< apacit( - hannel Elements
Hard Capacity
C 3ach "earer reuires
hardware in the $orm o$
channel elements 6Ces7 in
the #ode =! The num"er o$
C3s reuired depends on
the data rate% $or e9ample'
C >oice' , C3
C +5 ;"ps' 3 C3s
C ,2? ;"ps' * C3s
C 3?5 ;"ps' ? C3s
;/ard< apacit( - hannel Elements ;/ard< apacit( - hannel Elements
Hard Capacity
C Allocation is shared across
cells and carriers!
C (rovision must "e made to
accommodate so$t hand
over!
C 1o$ter hand over imposes
no additional "urden!
55) %i4ed network capacit(
$t is crucial that there is suBcient capacity in the 1xed network to
service the data 7ow" This data 7ow consists of control and
signalling channels as well as user data" $n particular the following
questions should be asked#
$s there suBcient capacity between the 3ode 9s and the 03(
+the $ub capacity,E
$s there suBcient capacity within the 03( itselfE
6oes the interface between the 03( and the M(5K3 +$u,
have suBcient capacityE
$s there suBcient capacity on the 03(-03( +$ur, linksE
6oes the core network have suBcient capacityE
The capacity on the interfaces is measured in terms of &8C links'
+approximately D Mbps," The fact that a 3ode 9 will generate more
traBc than a KM site means that more 1xed network capacity will
be required" $n KM networks, a single 8C link was suBcient to
serve the traBc from C< T0Ms" 2 site having more than C< T0Ms is
very rare" $n UMT networks, a single 8C link will not be suBcient
to service the traBc from a 3ode 9" $n areas where the demand is
low, it may be possible to get away with a single 8C link but it must
be borne in mind that hard blocking, rather than air-interface &soft'
blocking will be the limiting factor" The fact that the highest data
rate, packet switched, services will be available in the downlink only
means that it may well be necessary to have a greater provision
through the 1xed network in the downlink" This breaks with the
tradition of having the same capacity in both directions"
The $u interface between the 03( and the M( or K3 will have to
carry all the traBc from users of its 3ode 9s +with the exception of
that traBc that is for a user on the same 03(," $ts required
capacity will depend on the number of 3ode 9s that it serves"
The $ur interface allows 03(s to communicate with each other" This
is done, for example, when a U8 is in soft hand over between cells
that are controlled by di%erent 03(s" The &drift 03(' passes data
through to the &serving 03(' to allow decisions to be made
regarding active set membership and also to allow macro-diversity
gains to be realised" 4rovision of such a link is vital but the capacity
requirements are not expected to be large"
;/ard< apacit( - %i4ed Network ;/ard< apacit( - %i4ed Network
Hard Capacity
C Inter$aces must "e
su$$iciently provisioned!
C To #ode =s
C To the M1C)1-1#
C To other R#Cs
R#C
CIu to M1C)1-1#
CIu" to #ode =
CIur to other R#Cs
;/ard< apacit( - %i4ed Network ;/ard< apacit( - %i4ed Network
Hard Capacity
C The Iu" inter$ace
C The D3,F lin; is standard 62
M"it)s7!
C This may cause hard
"loc;ing as a #ode = air
inter$ace 63 cells7 can
support a total greater than
2 M"it)s!
R#C
CIu to M1C)1-1#
C"u0 to Node :
CIur to other R#Cs
;/ard< apacit( - %i4ed Network ;/ard< apacit( - %i4ed Network
Hard Capacity
C The Iur inter$ace
C (rovides a lin; to other
R#Cs where a @3 is in so$t
hand over with a cell in a
di$$erent R#C area!
C Low capacity!
R#C
CIu to M1C)1-1#
CIu" to #ode =
C"ur to other *Ns
;/ard< apacit( - %i4ed Network ;/ard< apacit( - %i4ed Network
Hard Capacity
C The Iu inter$ace
C .andles almost all user
data $rom all #ode =s!
C .igh capacity
reuirements!
R#C
C"u to MS9SASN
CIu" to #ode =
CIur to other R#Cs
55- /i'h Speed 3ownlink Packet .ccess 7/S3P.8
KM networks did not use spectrum in an eBcient manner because
the throughput was con1ned to a rigid structure based on timeslots
and raw bit rates" 3o matter how good the radio channel, the
throughput was limited to a raw rate of DN@ kbps" The introduction
of 86K8 allowed this theoretical rate to be trebled if the radio
environment could support it +i"e" (5$ was high J there was suBcient
&soft' capacity, by utilising an ;4* modulation scheme"
The 7exible way in which UMT cells allow users to share air
interface capacity makes it more eBcient" Aowever, there is still
the limitation imposed by the chip rate of !;>@ kcps and the O4*
modulation scheme" uppose a cell has one !;> kbps bearer
provisioned on the downlink" This bearer will have a maximum
power allocated to it that will allow service only if the radio
environment is of suBcient quality" 9ut, even if the radio
environment is of superb quality, it will not be possible to increase
the throughput" Aowever if A642 C?O2M is implemented, it will
be possible, in areas where the radio environment is very good, to
download at N?; kbps" This will increase the overall network
capacity by enabling a faster download in areas of high quality
radio environment"
;/ard< apacit( - /S3P. ;/ard< apacit( - /S3P.
Hard Capacity
C @ltimate constraint is
the 3?50 ;cps chip
rate!
C 2(1E allows 2 "its per
sym"ol!
C .igh 1peed &ownlin;
(ac;et Access
6.1&(A7 uses ,+ 2AM
which allows 5 "its per
sym"ol!
C .ard limit is dou"led!
CPSD
!6 C.M
;/ard< apacit( - /S3P. ;/ard< apacit( - /S3P.
Hard Capacity
C .1&(A is not a
DmagicF solution!
C Inter$erence limitations
still e9ist!
C .1&(A only possi"le
in areas o$ low
inter$erence!
C Aggregate networ;
capacity should "e
increased!
/S3P.
Possi0le
No
/S3P.
6"n-0uildin' Solutions
65! "ntroduction
The deployment of node 9s with cell antenna located inside
buildings will probably be implemented at the initial launch stage"
2lthough these help to ensure that the coverage within these
buildings is acceptable, the main reason for implementing such in-
building cells is to provide extra network capacity when it is
predicted that the macro-cell layer will not be able to serve the
o%ered traBc" 2s an example, consider an oBce building of D@
storeys with D@@ people on each 7oor, a total of >@@@ people" 2
particular operator may have ;@@ subscribers within such a
building" $f the average traBc o%ered by a subscriber is D< m8 of
voice, that represents an average loading of D@ 8" .hen video
telephony +LT, and data traBc are considered, it is clear that the
o%ered traBc justi1es at least one cell in its own right" 2s a single
macro-cellHs coverage area may contain several such oBce
buildings, it is clear that the capacity of the macro-cell layer will be
insuBcient to accommodate the o%ered traBc in such areas"
6eploying such indoor cells using the same frequency as the
macro-cell layer will improve the spectrum eBciency of the network
and also make network operations +in particular, hand over, much
more straightforward" The indoor cells must be implemented so as
to provide coverage within the building whilst minimising
interference between the indoor cell and the macro-cell"
"n "n- -0uildin' Solutions 0uildin' Solutions
In!uilding solutions
C Cells with indoor antennas can help
with coverage pro"lems!
C Most importantly% they add to the
networ; capacity and serve an
indoor DhotspotF!
C 3gL 20 $loor% 200 people per $loor
65000 people7' ?00 su"scri"ers% 20
3rlangs o$$ered!
C I$ this is >T% this would "e typical $or
a sectored #ode =! A macro/cell
may contain several such "uildings!
C In/"uilding solutions can alleviate
macro/cell capacity pro"lems!
%reBuenc( allocation %reBuenc( allocation
In!uilding solutions
C Advantageous i$ same
$reuency as macro/cell layer
can "e used!
C 1pectral e$$iciency ma9imised
C .and over "etween indoor and
outdoor environment simpli$ied!
C Mutual inter$erence must "e
minimised whilst engineering
so$t hand over region!
65$ The inter+erence loop
Mutual interference reduces capacity in a network due to the fact
that a form of feedback loop is established" $f a particular cell
su%ers external interference, it will send a &power up' message to
its U8s" This will cause uplink interference to increase at the
neighbouring cells that, in turn, send &power up' messages to their
U8s and thus the process continues" The result is that each cell
su%ers extra noise rise through interference, thus limiting capacity"
The fact that some of the cells are indoors and the macro cell is out
of doors, there is a level of isolation a%orded that helps to break the
feedback loop" .ithin a cell layer, the &problem areas' are those at
the cell border where the path loss to two or more cells is nearly
equal" $t is in these areas where the U8s su%er from downlink
interference and also generate uplink interference, thus reducing
frequency re-use eBciency" The physical barrier presented by the
outside walls of the building in question allows this region to
become negligibly small" -or example, from just inside the building,
the path loss to the indoor cell can be C@ d9 lower than to the
macro-cell" Pust outside the building, the reverse can be true" Thus
mutual interference is low and frequency re-use eBciency high"
8%ectively, it is a similar situation to that where no users are
located at the border region between cells"
; ;*educin' Mutual *educin' Mutual "nte+erence "nte+erence< <
C The lower the inter$erence the higher the capacity!
C =ecause o$ the single $reuency used in a @MT1 layer% there is
an DInter$erence $eed"ac; loopF!
C This means that inter$erence% rather than Gust adding to the
"ac;ground noise level% consumes a proportion o$ the networ;
resource 6power on the &L% noise rise on the @L7!
In!uilding solutions
*educin' Mutual *educin' Mutual "nte+erence "nte+erence
C The walls o$ the "uilding will help
provide isolation "etween the
indoor and outdoor cells% thus
improving capacity!
C D(ro"lem areasF are those where
the path loss to "oth cells is
similar!
C (resence o$ walls ma;es it
possi"le to ma;e this region
negligi"ly small!
C 1imilar% in principle% to a macro/cell
structure with gaps in coverage H
low inter$erence 6"ut .O $ailures H
"ut people donJt wal; through
walls7!
In!uilding solutions
65) The 3ead-Eone e++ect
$nterference will occur between operators, particularly where the
carriers are adjacent to each other" 2n in-building solution provides
very low path losses to the serving cell" This means that there is
the potential for very high interference values" (onsider a user who
is a subscriber to an operator that does not have a pico cell inside a
particular building" The pilot from the macro cell that serves the U8
may be as low as -C@< d9m" 4rotection for adjacent carrier
interference by approximately !! d9 can be assumed" $f a total
interfering level of -<N d9m was received at the adjacent carrier,
that would be reduced by the isolation 1lters to -G@ d9m giving an
8c5$o of -C< d9# the lowest level useable" 2ny larger interference
levels than -<N d9m would prevent the U8 from synchronising to its
serving pilot" $f the transmitting power of the pico cell was !! d9m
with a N d9i antenna gain, a path loss of anything less than GN d9
would cause problems" 2reas where the path loss is less than GN
d9 are expected to occur readily within pico cell coverage areas"
Two possible solutions to this problem may be considered#
2ll operators provide a pico cell solution within the building in
question +this would lead to all operators having identical
strategies for in-building deployment,"
)perators agree to share pico cell capacity and allow each
othersH users to hand over to the pico cell"
The 3ead Fone E++ect The 3ead Fone E++ect
C AdGacent carrier inter$erence may
"e particularly noticea"le within
pico cell environments!
C 33 d= is typical o$ the isolation
"etween adGacent carriers!
C Case shown illustrates the limiting
case where 3c)Io K /,* d=!
C Ma9 3IR( $rom pico cell is typically
50 d=m 633 d=m T9 (owerL 0 d=i
antenna7!
C (ro"lem cases will occur!
In!uilding solutions
CMacro cell pilot'
/,0* d=m
(ico cell
inter$erence' /*0 d=m
The 3ead Fone E++ect6 solutionsG The 3ead Fone E++ect6 solutionsG
C @nli;ely to provide a solution "y
engineering the radio environment'
low path loss to "est server is
generally a Dgood thingF
C (ossi"ilities
C All a$$ected operators deploy a pico
cell within a particular "uilding!
C Operators allow hand over to pico
cell carriers $rom a$$ected cells!
In!uilding solutions
CMacro cell pilot'
/,0* d=m
(ico cell
inter$erence' /*0 d=m
65- /and over +rom indoor to outdoor
The sudden di%erence in path loss between cells, although useful in
increasing frequency re-use eBciency, can cause problems with
hand over" $t is important that there is a &soft hand over' region
within which a U8 will have a connection to two or more cells" This
is the &border region'" $f a U8 suddenly receives a lot of power from
a cell that is not part of the active set, a call can drop" The U8
needs to be able to monitor this neighbour and execute the hand
over" $f the power from the neighbour increases too rapidly, there
could be insuBcient time to do this" Thus, the situation where the
border region coincides with the outside building wall is probably
going to result in problems with hand over" The indoor cell must be
engineered so that the border region obeys two conditions#
$t must be of suBcient si/e to allow soft hand over to take
place
The subscriber density within this border region should be low
/and Over /and Over
C A hand over region needs to "e
provided!
C 1udden changes in signal level
$rom a cell can lead to calls "eing
dropped!
C Reuired hand over region is near
the door!
C .and over region'
C Large enough to allow hand over!
C 1hould "e where su"scri"er
density is low% as hand over region
is area where mutual inter$erence
is highest!
C (reventing sudden changes in
signal strength at the .O region
reuires appropriate siting o$ pico/
cell and macro/cell antennas!
In!uilding solutions
Reuired hand
over region

65-5! En'ineerin' the 0order re'ion
Taking the above two constraints into consideration it is best if the
border region is designed to be just outside the building" This
entails designing the indoor antenna arrangement so as to make
the contour of equal path loss +to macro-cell and in-building cell,
enclose the building" $n this way, all the users within the building
will connect to the in-building cell only" -urther, at all points within
the building, the path loss to the in-building cell will be many d9
+typically C< d9, lower than the path loss to the macro-cell" This
will ensure that the users within the building will neither su%er
from, nor contribute to, interference with the macro-cell" $n the 1rst
instance, it is more straightforward if the con1guration of the in-
building cell +pilot power, maximum total power etc", is the same as
that of the macro-cell" Aowever, it is acknowledged that there are
cost and convenience advantages to be obtained if the in building
cell is of a lower power rating than the macro-cell"
The result of reducing the in-building cell power by, say, ? d9 is to
displace the hand over region from the region of equal path loss" 2t
the equal pilot strength region, the path loss to the macro-cell will
be greater than that to the in-building cell" Pust outside the soft
hand over region, in the area served by the macro-cell, the path
loss to the in-building cell could be lower than to the macro-cell"
Thus the in-building cell will su%er from high levels of uplink
interference because the U8 power is controlled by the macro-cell"
The seriousness of this problem will depend on the user density
within the a%ected areas" 2 less serious problem resulting from the
use of di%erent pilot powers is that the gain from soft hand over on
the uplink will be less than if the hand over region was located
where the path losses were equal"
/and Over /and Over
C Contour o$ eual pilot strength
6"etween macro/cells and pico/cell7
should "e engineered to "e Gust
outside% rather than Gust inside the
"uilding!
C This is "ecause su"scri"er density is
li;ely to "e much higher inside the
"uilding!
C This needs to "e chec;ed "y
measurements!
C This will lead to the path loss to the
pico/cell% $or su"scri"ers within the
"uilding% "eing much less than that to
the macro/cell!
C This is a good thing as it means the
pico/cell will have a negligi"le impact
on macro/cell capacity!
In!uilding solutions
Contour o$ eual
pilot strength!
Pico Pico- -cell Parameters cell Parameters
C (lanning and engineering is simpli$ied
"y using the same maGor parameters
6ma9imum power% pilot power% noise
rise limit7 are used $or the macro/cell!
C .owever% cost savings are to "e made
i$ the ma9imum power is reduced! The
pilot power would then have to "e
reduced!
C This will shi$t the hand over region
away $rom the area o$ ma9imum .O
gain!
C &isaster scenario is i$ the @3 hands
over to the macro cell while in the
"uilding!
C (ath loss to macro/cell could "e higher
than to pico/cell and hence @3 T9
power will increase! This has severe
implications $or the pico/cell capacity
In!uilding solutions
@3 could "e $orced onto
macro/cell! @3 power
would increase and pico/
cell would su$$er!
Pico Pico- -cell Parameters cell Parameters
C The pro"lem o$ the pico/cell su$$ering
$rom uplin; inter$erence does not
disappear i$ the hand over region is
designed to "e outside the "uilding!
C It is the su"scri"er density at the hand
over region that is important!
C (ico/cell capacity can "e improved "y
increasing its noise rise limit! This
could "e considered a sensi"le step in
conGunction with a reduction in pilot
power!
In!uilding solutions
.and over region in area
o$ low su"scri"er density
655 "mplementin' the in-0uildin' cell
The successful implementation of an &in-building solution' is an
area of expertise in its own right" The best approach will depend
upon the particular characteristics of the building" The number and
nature of the internal walls and 7oors is a major factor" The
di%erent methods commonly deployed are summarised and
compared below"
"mplementin' the in "mplementin' the in- -0uildin' solution 0uildin' solution
C &esign and implementation o$ in/
"uilding solutions is an area o$
e9pertise in its own right!
C The $ollowing provides an outline o$ the
decisions and choices regarding the
design o$ the pico/cell!
C A high level overview o$ the relative
advantages o$ the di$$erent options is
provided!
In!uilding solutions
6555! hoice o+ Node :
The major competing options here are#
2 central 3ode 9
Multiple compact node 9s
Utilisation of repeaters
The hoices The hoices 2 2 The Node : The Node :
C 1ingle #ode =
In!uilding solutions
C Multiple Compact
#ode =s
C Repeater to
e9ternal #ode =
6555!5! entral Node :
$n this case, a similar 3ode-9 to those used for outdoor sites
supplies the UMT signal to the whole building" There are the
following advantages and disadvantages of this solution"
ADVANTAGES
DISADVANTAGES
%llo"s easy capacity e#pansion (ne"
rac&s can be added to t'e e#isting
cabinet)
eing a centralised syste!( t'is
sol)tion can be a**ected by *a)lts
No need to 'a+e e#tra training *or t'e
tec'nical personnel,
-pace is re.)ired to locate t'e base
station
6555!5$ Multiple compact indoor Node :s
This solution consists in installing a certain number of compact
indoor 3ode-9Hs in the area to be covered"
ADVANTAGES
DISADVANTAGES
eing a distrib)ted syste!( t'is
sol)tion is rob)st to *a)lts,
Capacity e#pansion can be
proble!atic d)e to t'e installation o*
e#tra base stations,
-pace is not a critic re.)ire!ent d)e
to t'e s!all di!ensions o* t'ese
indoor Node/0s,
E#tra training needs to be done *or
t'e tec'nical personnel,
etter Co+erage can be pro+ided,
6555!5) *epeaters
0epeaters are very useful for enhancing coverage where no extra
capacity is required" They can be used to extend coverage from
either an outdoor cell site or to distribute coverage from an indoor
3ode-9Hs throughout the all area of interest" 0epeaters are two port
devices for direct connection to the base station and to an antenna
or a leaky cable" They can be connected to the base station by
means of a radio link +usually a Fagi directed at the base station, or
a cable in case a radio link cannot be established" -urther, it is
possible to establish a 1bre optic connection from the base station
to the repeater site and send the 0- signal over optical 1bre"
The hoices The hoices 2 2 The Node : The Node :
C 1ingle #ode =
In!uilding solutions
C 3asy capacity e9pansion 6Gust
add more cards7!
C 1ta$$ $amiliar with euipment%
as #ode = can "e the same as
$or the macro/cell!
C Centralised system could "e
prone to $aults!
C #ode = could "e physically
large to accommodate!
The hoices The hoices 2 2 The Node : The Node :
In!uilding solutions
C Multiple Compact
#ode =s
C More ro"ust to $aults as it is
distri"uted!
C Can lead to superior coverage!
C 1maller physical siAe!
C Capacity e9pansion can "e
harder!
C 1ta$$ may need e9tra training
on new type o$ #ode =!
The hoices The hoices 2 2 The Node : The Node :
In!uilding solutions
C Repeater to
e9ternal #ode =
C Cheap H "ut no solution to
networ; capacity pro"lems!
6555$ 3istri0ution methods
8ven if multiple 3ode 9s are used to provide the indoor solution,
each node 9 will usually be required to serve several 7oors" )nly
rarely will a single, centralised antenna be capable of providing
coverage for an entire 3ode 9 area" Usually, some kind of
distributed system needs to be implemented" There are a variety
of solutions put forward by di%erent vendors"
6555$5! . distri0uted antenna s(stem
This is a system comprising two or more antennas together with
associated power dividers and cabling" 8ach antenna would be
responsible for providing coverage in a particular /one"

ADVANTAGES
DISADVANTAGES
%llo" )se o* lo" radiated po"er, 1ig' total syste! losses in larger
b)ildings,
%llo" a *le#ible net"or& design 2
di**erent antenna !odels can be
installed in di**erent b)ilding areas,
Co!ple# cable installations
3ell s)ited *or se+eral b)ilding
str)ct)res
4is)al i!pact

6555$5$ 3istri0uted active antenna s(stems
6istributed active antenna systems represent an alternative to
systems using passive antennas" The antennas will contain their
own power ampli1ers" 6i%erent solutions of this type distribute the
signal from the central source by means of either optical 1bre or
(2T-< cable" The radiated power is usually low +e"g" C@@ m.,, thus
limiting the range" Aowever, as the antenna ampli1es the signal
itself, internal cable losses are not a signi1cant factor"
ADVANTAGES
DISADVANTAGES
4ery easy installation *or t'e optical
*ibre/C%56 cable,
7o" radiated po"er,
Red)ced E$C proble!s, Red)ced reliability and need o*
specialised personnel *or e.)ip!ent
installation and !aintenance,
7ess atten)ation losses t'an coa#ial
cables,
Po"er s)pply !ay be re.)ired at
eac' acti+e antenna site,
6555) *adiatin' ca0les
2n alternative signal-distribution method is represented by
radiating cables or &:eaky -eeders'" The radiating cable is acting as
continuous longitudinal antenna distributing the signal along its
path" This solution is particularly suitable for long structures such as
tunnels" The need for precision installation and the cost makes
them less suitable for oBce-type buildings"
ADVANTAGES
DISADVANTAGES
3ell s)ited *or longit)dinal str)ct)res, Not "ell s)ited *or general s'ape
str)ct)res (s.)ared or circ)lar),
8ni*or! signal distrib)tion along its
pat',
Not cost e**ecti+e in e+ery sit)ation,
7o" +is)al i!pact (can be 'idden
be'ind a s)spended ceiling),
Precision installation re.)ired,
The hoices The hoices 2 2 .ntenna S(stems .ntenna S(stems
C &istri"uted (assive
Antennas
In!uilding solutions
C &istri"uted Active
Antenna 1ystem
C Radiating Ca"le
The hoices The hoices 2 2 .ntenna S(stems .ntenna S(stems
C &istri"uted (assive
Antennas
In!uilding solutions
C Closeness o$ antenna to @3
allows low radiated power!
C Most suita"le antenna can "e
chosen $or each location
allowing good control o$
radiation!
C Ca"les can "e lossy and
e9pensive!
C Installation o$ heavy ca"le can
"e di$$icult!
The hoices The hoices 2 2 .ntenna S(stems .ntenna S(stems
In!uilding solutions
C Low grade CAT * ca"le can "e
used instead o$ $eeder ma;ing
installation easier!
C Lower $eeder losses
C Restricted to low transmit
power!
C (ower supply reuired at each
antenna location! C &istri"uted Active
Antenna 1ystem
The hoices The hoices 2 2 .ntenna S(stems .ntenna S(stems
In!uilding solutions
C 3asiest to model)predict lin;
loss!
C (roduces even coverage!
C Can "e hidden $rom view!
C O$ten the most e9pensive
solution!
C #ot suited $or all shapes
6"etter $or longitudinal shapes7!
C (recision installation reuired
to ma9imise "ene$its!
C Radiating Ca"le
6555- %ield measurements to veri+( the implementation
The requirement of the indoor cell is to provide coverage and
capacity inside the building whilst minimising the impact on the
macro-cell" -urther, hand over between the indoor and outdoor
environments should be possible"
The load on the macro-cell layer air interface will be minimal if the
path loss to the indoor cell is at least C@ d9 less than the path loss
to the macro-cell at points of interest" The &points of interest' are
the areas within the building where the user density is high J this
may be the entire building space" This can be ascertained by
measuring and comparing the pilot strengths of the macro-cell and
indoor cell at these points" $f the pilot powers are the same +a
useful starting point,, then the pilot strength of the indoor cell
should be at least C@ d9 above that of the macro cell pilots"
uccessful hand over depends on the existence of a soft hand over
region existing between the coverage areas" This region must be
large enough so as to ensure that the signal strength changes are
not too rapid for the hand over to be executed" These regions must
be established particularly at the entrances to the building"
-ortunately, users are not expected to be travelling at great speed
in these areas and a region of C@ metres width should be suBcient"
This can be either just inside or just outside the building, or both" $f
antenna adjustment is necessary to engineer the soft hand over
region, it is usually easier to adjust the in-building con1guration so
that its signal leaks slightly out of the entrance" $t must be
remembered however, that too much leakage will lead to mutual
interference between the indoor and outdoor cells"
%ield Measurements to heck on "mplementation %ield Measurements to heck on "mplementation
In!uilding solutions
C 1canner measurements can "e used to assess'
C Coverage
C .and over region
C Isolation $rom macro/cell! Ideally the macro/cell and the
pico/cell should not inter$ere with each other!
%ield Measurements to heck on "mplementation %ield Measurements to heck on "mplementation
In!uilding solutions
C Coverage
C (ico/cell pilot should "e "etter than appro9imately /,00
d=m at all locations 6note that there is no need $or an
Din/"uilding allowanceF such as $or outdoor drive test
measurements7!
C .and Over
C .and over region is near the "uilding entrance!
C @3s not e9pected to "e moving at high speed!
C ,0 m so$t hand over region should su$$ice!
C This can "e Gust inside or Gust outside the "uilding 6or
"oth7!
C 3asier to adGust in/"uilding antennas rather than macro/
cell antennas!
C Too much lea;age outside o$ the "uilding will result in
inter$erence!
%ield Measurements to heck on "mplementation %ield Measurements to heck on "mplementation
In!uilding solutions
C Isolation $rom macro/cell
C Loss to pico/cell should "e ,0 d= less than that to
macro/cell at all Dsigni$icant locationsF within the
"uilding!
C This can "e chec;ed "y comparing pilot strengths'
remem"ering to consider any di$$erence in pilot transmit
power and cell con$igurations 6M.A on macro/cell
would need to "e considered7!
#Usin' Micro-cells to Service
/ot Spots
#5! "ntroduction
The Macro-cell layer is designed to provide continuous coverage to
a speci1ed level +such as LT indoor," The initial result will be a
UMT carrier layer that will service a particular traBc density"
4roviding an indoor solution as described in the previous section
can accommodate extra traBc from oBce environments"
imulations suggest that, if D@ d9 of penetration loss must be
accommodated, a cell range of <@@ metres would be typical" This
would result in a site density of approximately D sites per square
kilometre and o%ered traBc of approximately DD 8rlangs per site
would be accommodated" $f outdoor environments covered by a
particular cell generate a high level of o%ered traBc, that traBc will
experience high levels of blocking" 2 micro-cell may be an
attractive solution in such a case" $n this instance, the possibility of
deploying a micro-cell re-using the carrier that is used by the
macro-cell layer is investigated"
Micro Micro- -cell plannin' cell plannin'
C T(pical ran'e +or macro-cell +or >T is 500 metres in dense ur0an
environment5
C Site 3ensit( +or covera'e appro4imatel( $ sites9km
$
5
C apacit( then appro4imatel( $$ Erlan's o+ >T per site 7-- Erlan's per
km
$
85
C .n area o+H sa(H !00 0( !50 metres would 0e e4pected to 'enerate onl(
one Erlan' o+ tra++ic5
C "+ an area o+ this siEeH or smallerH 'enerates !5 Erlan's o+ tra++icH a micro-
cell can help to accommodate this o++ered tra++ic5
Macro cell layer
providing
continuous
coverage
Micro cells
serving hot
spots!
Microcell Planning
Micro Micro- -cell plannin'6 carrier re cell plannin'6 carrier re- -use use
C I$ the macro cell carrier can "e re/used'
C 1pectral e$$iciency is improved!
C .and over "etween micro and macro cells is easier!
Macro cell layer
providing
continuous
coverage
Micro cells
serving hot
spots!
Microcell Planning
#5$ Micro-cell and "n-0uildin' cells compared
$n-building solutions using the same carrier as the macro-cell have
been shown to be a viable method of serving areas of high
subscriber density" 2 signi1cant feature is allowing straightforward
integration into the network is the isolation between the inside and
outside of the building that is provided by the walls of the building
itself" This isolation results in low levels of mutual interference
and, hence, higher levels of frequency re-use eBciency and cell
capacity" .hen a micro-cell is deployed in an outdoor environment,
such as a pedestrian shopping area, no such natural isolation
exists" The risk of high levels of mutual interference, leading to
reduced capacity, is therefore signi1cant"
Micro Micro- -cell plannin'6 mutual inter+erence cell plannin'6 mutual inter+erence
C In the case o$ in/"uilding solutions% the "uilding walls $ormed a
"arrier against inter$erence!
C This made re/use o$ the macro cell carrier more straight$orward!
C Outdoor micro cells have no such "arrier!
C (otential $or more serious inter$erence issues% reducing
capacity gains
In/"uilding
solution' walls
$orm "arrier
against
inter$erence!
Micro cells H no
"arrier against
inter$erence!!
Microcell Planning
#5) The Theor( 0ehind the Micro-cell
(onsider a macro-cell that serves an area of perhaps D@@@@@ m
D
" $f
video telephony is seen as the benchmark service, such a cell could
support up to approximately CD simultaneous connections"
uppose that a hotspot exists that covers about C<@@@ m
D
" $f this
area had a normal traBc loading, the area of the hotspot would be
expected to contain only C active user" The situation where this
area needs to support many active users is envisaged" 2 hotspot
should be deployed such that the e%ect of these many users, as far
as the macro-cell is concerned, is equivalent to one user of the
macro-cell" 8%ectively, this means that the total power of the many
users must equal that of the one macro-cell user" $f the micro-cell is
to support CD users, then the power of each of these CD users must
be, on average, CC d9 less than the power required for the macro-
cell" This can be taken to mean that the path loss to the micro-cell
must be CC d9 less than the path loss to the macro-cell"
Micro Micro- -cell plannin'6 theor( cell plannin'6 theor(
C 1uppose an area within a macro cell could accommodate only ,
3rlang o$ o$$ered tra$$ic i$ the macro cell capacity was divided
eually on an area "asis!
C #ow consider the situation i$ this area was e9pected to generate ,2
3rlangs o$ o$$ered tra$$ic!
Area generates
,2 times the
De9pectedF
tra$$ic!
Microcell Planning
Micro Micro- -cell plannin'6 theor( cell plannin'6 theor(
C I$ carrier $reuency is to "e shared and no e9tra loading to "e
placed on macro cell'
C 3ach @3 should operate at ,),2
th
o$ the power that it would i$ it
connected to the macro cell!
C (ath loss to micro cell should "e ,, d= less than that to macro cell!
Area generates
,2 times the
De9pectedF
tra$$ic!
Microcell Planning
Micro Micro- -cell plannin'6 theor( cell plannin'6 theor(
C This can "e arranged with pilot power settings "ut there are
pro"lems'
C I$ pilot powers are le$t eual% then "order region will "e where path loss
is eual "etween micro cell and macro cell @L inter$erence on macro
cell results!
C I$ micro cell pilot is ,, d= less than that $or macro cell% @3s Gust outside
the micro cell "order will cause a lot o$ @L inter$erence on the micro
cell!
Microcell Planning
(ilot powers eual'
macro/cell a$$ected
"y @3s on "order!
Micro cell pilot reduced "y
,, d=! Micro cell a$$ected
"y @3s Gust outside "order!
#5)5! Pilot Power Settin's
This sounds perfectly achievable but there is a complication that
becomes apparent when the cell-selection procedure is considered"
$f pilot powers of the micro-cell and macro-cell are made equal
then, at the border, the path loss to the macro-cell and the micro-
cell would be equal and the U8 would transmit with the same power
no matter which cell is selected" $n this situation the only bene1t
would be the macro-diversity gain from soft hand over" 2n option
would be to reduce the pilot power of the micro-cell so that users
would only connect to the micro-cell if the U8 transmit power was
going to be signi1cantly lower than if a connection was made to the
macro-cell" This, however, leaves the possibility that a U8
connected to the macro-cell would interfere signi1cantly with the
micro-cell thus reducing its capacity"
Micro Micro- -cell plannin'6 theor( cell plannin'6 theor(
C This can "e arranged with pilot power settings "ut there are
pro"lems'
C I$ pilot powers are le$t eual% then "order region will "e where path loss
is eual "etween micro cell and macro cell @L inter$erence on macro
cell results!
C I$ micro cell pilot is ,, d= less than that $or macro cell% @3s Gust outside
the micro cell "order will cause a lot o$ @L inter$erence on the micro
cell!
Microcell Planning
(ilot powers eual'
macro/cell a$$ected
"y @3s on "order!
Micro cell pilot reduced "y
,, d=! Micro cell a$$ected
"y @3s Gust outside "order!
#5)5$ En'ineerin' the Micro-cell
The problems described above become signi1cant only if there are
a signi1cant number of users in the transition regions" 8ngineering
of the micro-cell can avoid this situation occurring" $n order to
e%ectively deploy a micro-cell at the same frequency as the macro-
cell layer, the following guidelines should prove to be useful"
The capture area of the micro-cell should extend beyond the
hotspot" $n that way, the number of users in the
border5transition region should be small"
The radiation pattern, height and tilt of the micro-cell antenna
should be such that the path loss to the micro-cell rapidly
increases with distance once the U8 leaves the hotspot area"
This makes the border5transition region physically small which
again reduces the probability of users occurring within the
region of high mutual interference"
$f the above recommendations are adhered to, then the noise rise
limit and pilot power of the micro-cell could, in the 1rst instance, be
set to the same as the macro-cell" $n this way, the micro-cell could
be hoped to provide another ten LT connections whilst producing a
negligible e%ect on the macro-cell"
En'ineerin' the Micro cell En'ineerin' the Micro cell
C The success o+ an( strate'( depends on the user 0ehaviour5
C .reas o+ hi'h mutual inter+erence are onl( pro0lematic i+ there
are lots o+ users5
C Need to en'ineer the micro cell accordin'l(
C Micro cell dominance area should e4ceed the hotspot area5
Microcell Planning
Area o$ dominance o$
micro cell should
e9ceed the hotspot
area!
En'ineerin' the Micro cell En'ineerin' the Micro cell
C *adiation pattern o+ micro cell antenna should ensure that
path loss rapidl( increases once outside the dominance area5
Microcell Planning
C "nitial settin' o+ pilot power and N* limit can 0e the same as
+or macro cell5 "deall(H should 0e possi0le to reduce 3& power
to )# d:m 7pilot power would then have to 0e reduced
proportionatel(85
En'ineerin' the Micro cell6 +ield En'ineerin' the Micro cell6 +ield
measurements measurements
C "+ the pilot powers are eBualH the 0order area is at locations o+
eBual path loss5
C Pilot stren'th o+ micro cell should 0e !0 d: 'reater than that
+rom macro cell at all areas o+ hi'h su0scri0er densit(5
C "+ transmit pilot power o+ micro cell is 6 d: less than macro cellH
then micro cell pilot can 0e Iust - d: 'reater5
C Jhen micro cell pilot is reducedH pro0lem comes +rom potential
uplink inter+erence +rom macro cell UEs Iust outside the 0order5
C Should 0e possi0le to raise the N* limit to help with this5 7E5'5
macro cell N* limit6 - d:K micro cell N* limit6 1 d:85
Microcell Planning
#5)5) Jhat can 'o wron'G
The biggest problem is mutual interference" The macro cell base
station will still &see' the U8s served by the micro cell" $t is vital
that they transmit at a lower power than the U8s served by the
macro cell, otherwise they would produce a high level of uplink
interference on the uplink" Thus the nature of the hot spot should
be that the users would experience a much lower path loss to the
micro cell than to the macro cell" This is to protect the macro cell
from uplink interference" $t is tempting to attempt to achieve this
by scaling the pilot power +making the pilot power of the micro cell
lower than the macro cell," Aowever, this could produce the
situation where a U8 served by the macro cell caused a lot of uplink
interference on the micro cell due to the fact that the path loss to
the micro cell was much lower" There is inevitably going to be an
area +just as there is in the macro cell layer, where U8s are
subjected to a lot of downlink interference and generate a lot of
uplink interference# the border area" $n planning a micro cell, user
behaviour must be considered" The border area should not contain
a high density of users"
Possi0le Pro0lems Possi0le Pro0lems
C "+ there are lots o+ users in the 0order areaH this will cause
inter+erence pro0lems5
C "+ the micro-cell pilot power is reduced it ma( su++er +rom
inter+erence as the UEs connected to the macro-cell will 0e
transmittin' with relativel( hi'h power5
Microcell Planning
@sers in the "order
area will cause and
e9perience
inter$erence
pro"lems!
#5)5- 3etectin' Pro0lem .reas5
$f the pilot strength of the macro cell and micro cell are kept equal,
the border area is the area where the path losses are equal" The
high subscriber density should be restricted to areas where the pilot
strength from the micro cell is C@ d9 or more greater than the pilot
from the macro cell" This can be checked by making 1eld
measurements with a scanner"
Usin' Scanner Measurements Usin' Scanner Measurements
C Jith pilot powers o+ the micro-cell and macro-cell set to eBual
levelsH the stren'th o+ the micro-cell pilot should 0e at least !0
d: 'reater than the macro-cell pilot throu'hout the area o+
hi'h e4pected user densit( 7the ;hotspot<85
Microcell Planning
Area o$ e9pected
high user density!
En'ineerin' the Micro cell6 +ield measurements En'ineerin' the Micro cell6 +ield measurements
C "+ the pilot powers are eBualH the 0order area is at locations o+
eBual path loss5
C Pilot stren'th o+ micro cell should 0e !0 d: 'reater than that
+rom macro cell at all areas o+ hi'h su0scri0er densit(5
C "+ transmit pilot power o+ micro cell is 6 d: less than macro cellH
then micro cell pilot can 0e Iust - d: 'reater5
C Jhen micro cell pilot is reducedH pro0lem comes +rom potential
uplink inter+erence +rom macro cell UEs Iust outside the 0order5
C Should 0e possi0le to raise the N* limit to help with this5 7E5'5
macro cell N* limit6 - d:K micro cell N* limit6 1 d:85
Microcell Planning
#5- /otspots straddlin' macro cell 0oundaries
o far we have considered the case where the hotspot lies totally
within an area served by a single macro cell" This is not necessarily
the case" $t may be common for a hotspot to lie on the border area"
$n fact, hotspots in such locations are likely to cause greater
problems with regard to macro cell layer capacity as the border
areas are those where mutual interference is highest" 2 micro cell
can provide a great bene1t in such circumstances" The principle of
engineering the micro cell is the same as in the case where the
hotspot lies within a macro cell coverage area" 3amely, the area of
high user density should experience a much lower path loss to the
micro cell than to the macro cell"
/otspots straddlin' cell 0oundaries /otspots straddlin' cell 0oundaries
C . hotspot straddlin' a cell 0oundar( will 0ene+it 'reatl( +rom a
micro cell as it is in an area that will cause si'ni+icant
inter+erence5
C Similar en'ineerin' considerations appl(6 the micro-cell
covera'e area should e4ceed the area o+ hi'h su0scri0er
densit(5
Microcell Planning
#55 Propa'ation modellin' +or micro cells
The initial planning of a UMT network is often conducted with map
data with a resolution of perhaps <@ metres" This is not a 1ne
enough resolution to simulate the e%ect of a micro cell" Map data
at a 1ne resolution should be used together with an appropriate
propagation model" Models used for macro cell planning rely on
empirical models that do not consider building re7ections explicitly"
9uilding re7ections, penetration and di%raction form the dominant
propagation mechanisms in micro cells" 2ccordingly, a &ray
tracing' model that considers these mechanisms should be used"
Prediction *esults Prediction *esults
C @sing O;umura/.ata% you will get predictions that are largely "ased on the
distance $rom the Cell!
C This has a certain validity when the antenna is a"ove the "uilding "ut
not when it is down "elow "uilding height!
Microcell Planning
Prediction *esults Prediction *esults
C :hen di$$raction and scatter are the main mechanisms% the $ield strength will
change in a much less straight$orward manner!
Low signal strength at
some locations close
to =T1!
DCanyon 3$$ectF
Microcell Planning
.daptations to Macro .daptations to Macro- -cell Model cell Model
C Most signi$icant di$$erences "etween Micro/cell and
Macro/cell predictions'
C Canyon 3$$ect' Low reduction in $ield strength with
increasing distance 6De9ponentF appro9imately 2!07!
C Modelled "y' adGusting appropriate parameters to
give low e9ponent!
C Coverage holes close to =T1!
C Modelled "y' more sophisticated macro/cell models!
Microcell Planning
Enhancement to Macro Enhancement to Macro- -ell Models ell Models
C Clutter O$$set in d=
C .eight and 1eparation' @sed $or &i$$raction
calculations 6time consuming7!
C Through loss in d=);m' Ma9imum distance
speci$ied! Loss is weighted $rom Aero to , over this
distance! 8aster than di$$raction calculations!
1eparation
Through/loss
distance
Microcell Planning
Enhancement to Macro Enhancement to Macro- -ell Models ell Models
C (lausi"le results 6statistically uite good regarding
s!d! o$ error7 can "e o"tained "y enhancing a macro/
cell model! .owever'
C DAlways strugglingF' model has to "e care$ully
adapted $or each situation! (arameter tuning
"ecomes something o$ a D$iddle $actorF! :ould
Dtuned modelF "e appropriate $or general useB
C #o Dset up and goF capa"ility!
C -reater accuracy is o"taina"le $rom deterministic
models!
Microcell Planning
)A Network con+i'urations )A Network con+i'urations
Macro/cell'
antenna well a"ove roo$tops
Micro/cell'
antenna "elow roo$tops
Mini/cell'
antenna on small
"uilding
roo$tops
6"elow higher "uildings7
Microcell Planning
Radio confgurations: Mini Radio confgurations: Mini
Q $ntermediate con1gurations#
C The emitting antenna is located on a roo$ top "ut not higher than the
surrounding roo$ level
C The emitting antenna is against a "uilding $aRade "ut not lower than the
surrounding roo$ level
C Mini/cellular con$iguration
C The radio energy is partly propagated a"ove
o"stacles and partly along the streets
C Choice must "e the >olcano Mini model
o, !6 channel modelling is required
Microcell Planning
Operational cases Operational cases

2ccurate coverage in rural hilly area
(6M2 network coverage +U2,
Multi-band KM Macro-Micro (ell
Mini-cell model for !K
Microcell Planning
Propa'ation Models Propa'ation Models
Pure deterministic
models
Semi-statistical models
C
o
s
t
/
.
a
t
a
w
i
t
h
&
i
$
$
r
a
c
t
i
o
n

c
o
r
r
e
c
t
i
o
n
s
3eterministic models
>
o
l
c
a
n
o
/
&
e
y
g
o
u
t
Statistical models
C
o
s
t
/
.
a
t
a
>olcano/RT&
Microcell Planning
Jhich topo'raphic data0ase Jhich topo'raphic data0aseG G
C The right data"ase $or the right environment' trade/o$$ calculation time vs!
accuracy
C &ata"ase uality' adapted to telecom pro"lems
C Importance o$ the clutter description
C The good choices'
C &eterministic vs! statistical
C .igh vs! low resolution data"ases
*eBuired la(ers
Environment
3 rasters 62 H * m7 4
, vector layer
Or 3& vectors
3 rasters )
vectors
6* H ,0 m7
2)3 rasters
6,0 H *0 m7
Ur0an Su0ur0an *ural
C m
D@ m
Microcell Planning
3e('out 3e('out model model

E
9
P
d
1
d
2
h
v h
d
d

2 1 2
1 2

d

+ d
( )
if v
A v u u
u v
> -0.7 (cas NLOS)
( ) . !" + +

# $ 20 1
1
%(+)
+
0
/:d
5errain pro*ile
Microcell Planning
*a( *a(- -tracin' tracin'
Ray contri"utions constructed $or an emitting
site located against a "uilding wall
Microcell Planning
. deterministic propa'ation tool6 . deterministic propa'ation tool6
e45 >olcano *T3 Model e45 >olcano *T3 Model
Only &eygout
Only RT)@T&
>olcano
3 3
R R
4 4
/oriEontal plane
*a( Tracin'
Microcell Planning
ali0ration 9 measurements ali0ration 9 measurements
6distance7 Log
Measures
Microcell Planning
C #ot mandatory "ut recommended
C 3nvironment not entirely descri"ed "y &TM 6stores% "alconies7
C Automatic cali"ration
C 1imple and uic; cali"ration
C S8ree 1paceT Correction
C :eighting
C 8ull cali"ration
C #on linear parameters 6"uilding heights7
1tandard deviationK * to ? d=
Macro Macro- -cellular cellular covera'e covera'e
C 1hadowing prediction
C @p to long distance realistic prediction
CO1T H .ata 1imulation 6tuned7 >olcano 1imulation
.igher terrain
1treet
.ill
Losses 6d=7
Microcell Planning
Micro Micro- -cellular cellular covera'e covera'e
C 1hadowing and wave/guiding e$$ects
C @p to long distance realistic prediction
1tatistic simulation 6.arley model7 >olcano 1imulation
Received
power 6d=m7
Microcell Planning
New important +eatures New important +eatures
(overage Map at C"<m
(overage Map at N"<m
(overage Map at C!"<m
Multi7oor coverage for mini-cellular site
Microcell Planning
An important parameter An important parameter
Downlink orthogonality factor Downlink orthogonality factor
Aow is this 6)- has been taken into account so farE
Klobally, 4er 2rea, 4er (lutter
2 new solution# by Lolcano deterministic simulations
)utdoor only )utdoor and indoor
Microcell Planning
#56 Multiple micro cells
$t is possible that a hotspot is physically too large to be served by a
single micro cell without border problems arising" $n such cases, a
small cluster of micro cells can be established" Throughout the
hotspot served by this cluster, the path loss to the micro cell should
be much lower than to the macro cell" 2gain, this means that the
hotspot should be geographically distant from the macro cell" $t is
not possible to establish an additional continuous cell layer,
consisting of micro cells, that shares the frequency of the macro cell
layer" 2 micro cell layer requires the use of a second carrier
frequency"
#5# &imitin' +actors
6eploying micro-cells that use the same frequency as the macro-
cell layer is clearly a powerful solution to capacity problems" $t is
not, however, without adverse consequences" $f we consider a
problem that has resulted in multiple micro cells being deployed, it
must be remembered that the reason for deployment was the high
subscriber density in that area" 3earby macro-cells will inevitably
su%er from greater uplink interference produced by these users,
thus reducing the capacity of these cells" This can be limited by
optimal engineering of the micro-cells but there is a limit" Much
greater isolation between macro-cells and micro-cells is achieved if
they are allocated separate frequencies" This is discussed later in
the section on hierarchical cell structures"
1The E++ect o+ %urther Site
Sectorisation
15! The sectored antenna
)mni-directional sites are now almost unheard of in an urban
environment" The fact that sectored sites will service a greater
subscriber density is well established" The standard level of
sectorisation is three sectors per site" 2ntennas have been
developed with this level of sectorisation in mind" 2lthough each
sector will have to control an angle of CD@, the arrangement of
sites means that the required range reduces as you move away
from the principal direction" This leads to antennas with ;< degrees
or, more recently, ?< degrees beamwidth being adopted" $t is found
that the ?<-degree beamwidth antenna is more e%ective at limiting
mutual interference"
%urther Sectorisation o+ Sites %urther Sectorisation o+ Sites
Furt"er Sectorisation
C Three/sectored sites
have evolved to "e the
norm in ur"an and
su"ur"an areas!
C 3ach antenna controls a
,20U sector!
C Antenna "eamwidth a
compromise "etween
coverage and
inter$erence!
C +*U is the most common!
C ,? d=i is typical gain!
15$ "ncreasin' the level o+ sectorisation
2dding further sectors to a site could allow a greater user density to
be served" Aowever, any increase can be limited by mutual
interference" Kenerally, the greater the cell density, the higher the
interference levels"
0emembering that throughput
) 1 (
3840
0
i N E
b
+


, the value of i is
signi1cant"
2 speci1c antenna, with a !< beamwidth, has been developed for
sites that are con1gured with six sectors, at ?@ intervals"
%urther Sectorisation o+ Sites %urther Sectorisation o+ Sites
Furt"er Sectorisation
C 1i9/sectored sites could% in
theory% dou"le the capacity!
C I$ mutual inter$erence
increases% capacity may not
increase as e9pected!
C =ut% i$ antennas are higher
gain% can increase!
( )

i
N
E
b
+

1
3840
Capacity
0
%urther Sectorisation o+ Sites %urther Sectorisation o+ Sites
Furt"er Sectorisation
C Antenna "eamwidth is
highly signi$icant in arriving
at the optimum "etween
coverage and inter$erence!
C 3*U is seen as the most
appropriate!
C 2, d=i is typical gain!
C Monte Carlo simulations
can uanti$y the li;ely
improvement!
15) Usin' simulations to assess the e++ectiveness
ix-sector sites are almost certainly going to be an upgrade in
UMT networks" 2s a test of their e%ectiveness, the same area is
assessed for capacity before and after the upgrade"
15)5! >ital statistics6
4arameter ! sector site ? sector site
2ntenna
9eamwidth
?< degrees !< degrees
2ntenna Kain C;"! d9i DC"@ d9i
30 limit > d9 ? d9
9enchmark ervice ?> kbit5s LT ?> kbit5s LT
Target 8b53o ! d9 ! d9
ites were placed such that the maximum coverage range was <@@
metres" This was suBcient to provide continuous indoor coverage
with a building penetration loss of D@ d9" 2n area of < km by ! km
was covered using !! sites"
15)5$ Estimates o+ capacit(
The pole capacity of a cell with no interference with an 8b53o of !
d9 is approximately CGD@ kbit5s" $n the three-sector case, the noise
rise limit of > d9, represents a ?@= loading factor# CC<D kbit5s" This
represents approximately C; simultaneous connections that will
serve CC"< 8rlangs of o%ered traBc" 2n out of cell interference
ratio of @"? will reduce this to N"D 8rlangs per cell" The < by ! km
area was spread with ND@ 8rlangs of o%ered traBc and a simulation
was conducted with the U8s being placed indoors" 2s expected,
satisfactory results were obtained" The network was seen to be
operating near its limits, however" The o%ered traBc was then
doubled to C>>@ 8rlangs" The result was that the network reached
saturation at a level of approximately G<@ terminals being served
+??= success rate, before blocking occurred" $n this situation the
major cause of blocking was downlink capacity +the required 8b53o
on the downlink is expected to be approximately N d9, but there
were a signi1cant number of failures on the uplink with the noise
rise limit being hit frequently" 2t this stage, the e%ect of further
sectorisation was simulated" 2dding new cells with new power
ampli1ers +so each site had six cells of similar con1guration to the
original three cells, resulted in the C>>@ 8rlangs of traBc being well
served" 2n average blocking ratio of != and a success rate of G>=
of all attempts +note that a failure due to the path loss being too
high does not constitute blocking, were reported with the failures
split between the uplink and downlink"
$n order for a comparison to be made with equal power provided to
each site, the simulation was re-run with the power for each of the
six-sector cells being halved" This revealed a slight degradation so
that G@";= of all attempts were successful" $t was noticed that the
majority of failures were due to the limit of power per connection on
the downlink being reached" This was set at C@= of maximum total
power" $ncreasing this limit to C?= of total power resulted in an
improvement to the success rate to GD"D=" This meant that the six-
sector area was serving an average of C!!@ terminals compared
with G<@ in the three-sector situation" This is an increase of >@=
but with a much higher grade of service o%ered to the customer
+approximately C@= blocking compared with over ;@= blocking,"
$f the o%ered traBc in the six-sectored area was increased again to
cause saturation of the network, the six-sector area then served an
average of C;!@ terminals, an increase of more than G@= on the
three-sector situation"
2s an indication of the levels of mutual interference, the frequency
re-use eBciency +the percentage of U8 power received by a cell
that comes from users of that cell, was approximately ?!= for both
the !-sector and the ?-sector networks"
Thus it can be concluded that further sectorisation of sites
represents a powerful method of increasing the capacity of a
network"
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C An area o$ ,* ;m
2
was
planned to give +5 ;"it)s >T
service indoor!
C 33 sites% << cells with a
range o$ *00 metres were
needed!
C Target 3")#o ta;en to "e 3
d= on @L
C #R limit set to 5 d= 6+0M
loading7!
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C Capacity estimation'
C This would accommodate
,? simultaneous
connections!
C 3rlang = ta"le 62M
"loc;ing7 suggests this
should serve ,,!* 3rlangs!
C Inter$erence ratio o$ 0!+ will
reduce this to 0!2 3rlangs
per cell!
( )
&bps 1160
10
: , 0 3840
3 , 0

%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison


Furt"er Sectorisation
C << cell networ; should
support 020 3rlangs!
C 1imulation con$irms this'
C <+M call attempt success!
C 8ailures distri"uted "etween
@L and &L 6"oth capacity and
path loss $ailures noted7!
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C #e9tL the o$$ered tra$$ic was
dou"led to ,550 3rlangs!
C #etwor; is now saturated!
C +*M call attempt success
6<*0 connections on average7!
C ?*M "loc;ing!
C &L is particularly
overloaded 6note' higher
3")#o reuired on
downlin;7!
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C 3ach site then had a $urther
three sectors added!
C 3* degree antennas
deployed throughout!
C #oise Rise limit increased
to + d=!
C Result'
C <5M call attempt success!
C 3M "loc;ing!
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C #e9t' an Deual powerF
comparison!
C &L power halved to each
cell!
C Result'
C <0!?M call attempt success
6c!$! <5M7!
C 8ailures tend to "e on &L due
to "earer power limit "eing
reached!
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C #e9t' adGust "earer power
limit!
C (ower per connection
increased $rom ,0M o$
ma9imum power to ,+M o$
ma9imum power!
C Result'
C <2!2M call attempt success!
C 8ailures evenly distri"uted!
C Average connections
served' ,320 6c!$! <*0 with
3/sectored sites7!
%urther Sectorisation o+ Sites - omparison %urther Sectorisation o+ Sites - omparison
Furt"er Sectorisation
C 8inally' dou"le o$$ered tra$$ic
to saturate networ;!
C ,?30 terminals served on
average!
C <0M more than $or 3/
sectored networ;!
C Conclusion is that $urther
sectorisation o$ sites is a
potentially power$ul method
o$ increasing networ;
capacity!
15- Nei'h0our Plannin'
2dditional sectorisation can cause further problems with neighbour
planning" The list of neighbours will necessarily become longer
considering the following#
2ll co-sited cells should be declared as neighbours"
There are more cells within a given distance"
Aowever, the fact that the energy radiated from a cell is con1ned to
a narrow beam means that the area over which a cell causes
interference will be less" The overall e%ect of this is that the
neighbour list will probably become longer in the ?-sector case than
it was for the !-sector situation but not unbearably so" $f a
neighbour list of CD neighbours were seen as typical for a !-sector
area then perhaps C? neighbours would be typical for the ?-sector
area"
%urther Sectorisation o+ Sites - Nei'h0ours %urther Sectorisation o+ Sites - Nei'h0ours
Furt"er Sectorisation
C Eeeping the neigh"our
6#cell7 list short is seen as
good practice!
C 3ach cell will have more
neigh"ours i$ $urther
sectorisation is employed!
C Co/sited cells should "e made
neigh"ours!
C There will "e more cells within
a given range!
C .owever% energy $rom each
cell is con$ined to a narrow
"eam and will e$$ect a
smaller area!
%urther Sectorisation o+ Sites - Nei'h0ours %urther Sectorisation o+ Sites - Nei'h0ours
Furt"er Sectorisation
C Result is that the #cell list
will have to "e longer i$
$urther sectorisation is
employed% "ut not
un"eara"le so!
C I$ #cell list is typically ,2 $or
an area where 3/sector cells
are used% a list o$ ,+
neigh"ours should "e
su$$icient where +/sector
cells are the norm!
,Usin' .dditional arriers 2
/ierarchical ell Structures
,5! Spectrum .llocation
)perators are not allocated a single carrier" 0ather, a block of two
or three adjacent carriers is normally allocated" This allows for a
7exible approach to be taken to providing additional network
capacity" 8ssentially, this is equivalent to two or more networks
operating in parallel" Two issues require addressing#
Aand over between the two carriers is a &hard' hand over" Aard
hand overs involve the U8 entering compressed mode and
putting an extra burden on the network"
There will be interference between the adjacent carriers" $n
most cases, this is not expected to cause signi1cant problems
for a particular network +where rival networks operate at
adjacent frequencies, problems can exist" The section on $ndoor
olutions goes into detail on this issue,"
Usin' .dditional arriers Usin' .dditional arriers
Multicarrier deployment
C Issues to consider'
C .ard hand over "etween carriers' compressed
mode operation!
C AdGacent carrier inter$erence to "e considered!
,5$ 3eplo(in' e4tra carriers in the macro cell la(er
$t is possible to increase the capacity of the macro-cell layer by
deploying an additional carrier" This is particularly attractive if an
operator has ! carriers as it leaves the third carrier for use with
micro-cells and pico-cells"
,5$5! . test case
$n order to evaluate the likely improvement, the standard C< km
D

area with !! sites was considered +as in the case where further
sectorisation was evaluated J see previous section," 2 single
carrier, !-sectored network can support ND@ 8rlangs of LT" The
o%ered traBc was doubled to C>>@ 8rlangs and an extra carrier was
deployed" The simulation assumed that the two frequencies shared
an antenna and that the power was shared between the two
frequencies" 3evertheless, it was found that the capacity of the
network was very nearly doubled as a result of the deployment of
an extra carrier" -urther, interference between the carriers did not
have a noticeable e%ect on the predicted performance"
To evaluate the capability of a macro-cell layer to accommodate
very high traBc densities, the exercise was repeated with each site
divided into six, dual-frequency, sectors" Thus, each 3ode 9
contained CD UMT cells" This was found to support a further
doubling of traBc to DN@@ 8rlangs, a density of almost D@@ 8rlangs
of video telephony per km
D
" This is an estimate for a site density of
D"C sites per km
D
+thus each site supports almost C@@ 8rlangs of
video telephony traBc," $f this is to be increased further, extra
diversity could be used" 2s the simulations suggest that the
network is reasonably balanced +failures were evenly distributed
between U: and 6:, any enhancement would have to be
implemented on the uplink and the downlink" Thus, perhaps, four-
component diversity could be implemented on the uplink +instead
of the standard, single cross-polar diversity deployed as standard,
together with space diversity on the downlink" The downlink would
be the limiting factor in determining the likely capacity increase
that would result" 8stimates vary regarding the likely increase, with
!@= being a typical 1gure"
Aigher subscriber densities could be accommodated by increasing
the site density" 3ot only is this an expensive option, it is diBcult to
implement retrospectively" The original plan was to provide indoor
coverage at the cell edge" This determined the site density of the
macro-cell layer"
omparison usin' the simulator omparison usin' the simulator
Multicarrier deployment
C Initial situation'
C 1ingle carrier
C 33 sites 6<< cells7 in ,* ;m
2
C 020 3rlangs o$ >T carried
C 2 carrier situation'
C Two carriers deployed per cell
C (ower split "etween two carriers
C Capacity increased to ,3*0 3rlangs o$ >T
omparison usin' the simulator omparison usin' the simulator
Multicarrier deployment
C 2 carrier% + sector situation'
C Capacity $urther increased to 2000 3rlangs >T
C &ensity almost 200 3rlangs);m
2
with a site density o$ 2!,
sites);m
2
C 8ailures distri"uted evenly "etween uplin; and
downlin;!
C Any $urther increase 6through diversity etc!7 would have
to "e applied to "oth uplin; and downlin;! &ownlin;
diversity improvement e9pected to "e appro9imately
30M 6this would $orm the limiting $actor7!
,5) %i4ed network provisionin'5
$f each site can handle C@@ 8rlangs of LT, the peak loading would
be estimated at CC> connections" The aggregate user data rate
would be N"! Mbit5s" .ith an overhead for signalling and control,
this would increase to approximately G Mbit5s" 4rovisioning of this
through 8C links would be a signi1cant consideration"
This is four the case where each 3ode 9 controls CD cells" The
requirement would be less if the &?-cell' or &!-cell' option was
chosen"
3umber of
cells per
site
8rlangs
per site
4eak
connection
s per site
2ggregate
user bit
rate per
site
8stimated
requirement
! D< !> DCN?
kbit5s
DN@@ kbit5s
? <@ ?C !G@>
kbit5s
>G@@ kbit5s
CD C@@ CC> NDG? G@@@ kbit5s
kbit5s
The basic &building block' of 1xed network transmission is the &8C
link' that carries D@>; kbit5s" $t is tempting to provision a !-cell site
with a single 8C link, at least initially" $f this option is implemented,
it is the 1xed network that is likely to become the 1rst factor
limiting network capacity" -ixed network capacity must be
addressed before the air interface capacity is enhanced"
%i4ed network implications %i4ed network implications
C Capacity o$ ,003 per site would necessitate an
upgrade to the $i9ed networ;!
Multicarrier deployment
3umber of
cells per site
8rlangs
per site
4eak
connections
per site
2ggregate
user bit rate
per site
8stimated
requirement
! D< !> DCN? kbit5 s DN@@ kbit5 s
? <@ ?C !G@> kbit5 s >G@@ kbit5 s
CD C@@ CC> NDG? kbit5 s G@@@ kbit5 s

,5- arrier loadin' strate'(5
.hen multiple carriers are used on a network, it is possible to
decide on a priority of loading" .hen the two carriers operate in
parallel, as in the case above where the extra carrier is used to add
further capacity to the macro cell layer, it is best to load them
equally" $f one carrier is more heavily loaded than the other, the
8c53o on that carrier will be worse than it is on the lightly-used
carrier" This will lead to hard hand over attempts being made
between the two carriers" 2voiding these is a good thing"
2dditionally, the power levels required would be higher on the
carrier that had higher levels of activity" Kenerally, keeping power
levels to a minimum is seen as good practice in a UMT network
because, for example, it leads to the maximum coverage for higher
data rate services" (arrier loading strategies will be re-visited later
in this section following consideration of hierarchical cell structures"
arrier arrier- -loadin' Strate'( loadin' Strate'(
C I$ two carriers are applied to the macro/cell
layer% it is "ene$icial to load them eually!
C I$ one carrier is more heavily loaded% 3c)Io will "e
worse leading to hard hand over occurring!
C Reuired "earer powers will "e higher in more
heavily loaded carrier H minimising power is a
generally desira"le aim!
Multicarrier deployment
,55 /ierarchical cell structures5
.hen o%ered traBc grows beyond approximately C@@ 8rlangs per
km
D
, it is usually concentrated at &hotspots' rather than evenly
distributed across the coverage area" )ne obvious case is where
oBces generate a lot of traBc" $ndoor solutions usually alleviate
any problem associated with this" )ther areas that generate very
high subscriber densities include pedestrianised shopping areas
and open-air sports venues, such as racetracks"
The isolation provided by the walls of the building itself makes it
possible to implement an indoor solution using the same frequency
as the macro-cell layer" .ith an outdoor micro-cell, there is not so
much isolation" This will limit the e%ectiveness of a micro-cell
solution, as shown in section N" $f a separate carrier is used for the
micro-cell, this will provide isolation +in the frequency domain,
rather than physical isolation," That, in turn, should lead to greater
throughput being possible"
/ierarchical ell Structures /ierarchical ell Structures
C The su"scri"er density that can "e
accommodated "y a macro/cell layer will have a
limit 6,00 3rlangs);m
2
7!
C .igher densities is usually due to DhotspotsF that
can "e treated separately'
C O$$ice "uildings' in/"uilding solution reuired
C 8oot"all stadiums% railway stations etc!' micro/cell
can "e deployed!
Multicarrier deployment
/ierarchical ell Structures /ierarchical ell Structures
C The e$$ectiveness with
which a micro/cell can
"e deployed sharing the
same $reuency and the
macro/cell layer
depends upon the a"ility
to isolate the micro/cell
$rom the macro/cell!
C This is easier i$ the
micro/cell is at a
considera"le distance
$rom the macro/cell!
C The use o$ a separate
carrier $or the micro/cell
must "e considered!
Multicarrier deployment
CPossi0le to serve
hotspot with micro/cell
that re/uses macro/cell
$reuency
C3i++icult to serve
hotspot with micro/cell
that re/uses macro/cell
$reuency
/ierarchical ell Structures /ierarchical ell Structures
Multicarrier deployment
C The deployment o$ micro/cell can "e e9tended to provide a second layer!
C A separate carrier is essential!
C Capacity o$ micro/cell layer can "e dou"le that o$ the macro/cell layer!
,555! apacit( o+ micro-cells usin' separate carriers
The major factor that limits the capacity of any UMT cell is intra-
frequency interference" This is particularly the case in the downlink
direction as it counteracts the bene1cial e%ects of orthogonality" $f
micro-cells are deployed to service hotspots, and are allocated a
separate carrier frequency, it is possible to e%ectively isolate them
from such interference" -urther, if any multipath is over a short
extra path distance then the orthogonality of a micro-cell should be
much better than that of macro-cells" Lalues of orthogonality
around @"G are expected to be typical for a micro-cell whereas @"? is
typical for a macro-cell" This makes the air interface capacity
several megabits a second in the downlink direction"
,555$ Pilot and common channel powers in micro-cells
4ilot and common channel default settings are decided upon as
appropriate when cells are experiencing signi1cant levels of
external interference, such as in a macro-cell layer" .hen the
levels of interference are lower, the powers allocated to pilot and
common channels can be reduced" This in turn can further increase
the downlink capacity"
ommon and Pilot Powers in Micro ommon and Pilot Powers in Micro- -cells cells
C (ilot powers o$ appro9imately ,0M o$ ma9imum power
ensure that the pilot can "e used $or synchronisation in
places o$ ma9imum inter$erence considering
inter$erence levels and orthogonality $actors present in a
macro/cell layer!
C I$ levels o$ inter$erence are lower and orthogonality is
higher% the pilot and common channel powers can "e
reduced% perhaps to *M o$ ma9imum power!
Multicarrier deployment
,555) &ink 0ud'ets +or micro-cells
)rthogonality is improved in micro-cell environments" This is
because the path length di%erences are usually very small"
Aowever, the result of re7ections with small path length di%erences
is to increase the probability of 7at fading" $n certain circumstances
this will lead to such rapid changes in 1eld strength that user
mobility will lead to higher target 8b53o values" $ncreases in target
8b53o of perhaps D d9 can be expected but this will be
environmentally- and mobility-dependent"
&ink 0ud'et +or micro &ink 0ud'et +or micro- -cells cells
C &ue to high levels o$ multi/path with small path length
di$$erences% there will "e considera"le $lat $ading!
C A $lat $ading margin may "e reuired!
C 2 d= is suggested as an appropriate value!
C This will "e environmentally and mo"ility dependent!
Multicarrier deployment
,555- Multi-la(er strate'ies +or dense ur0an environments
The initial roll-out of a macro-cell layer will be dominated by
coverage concerns" $n order to provide LT coverage indoors, this
will lead to a site density of approximately two sites per square
kilometre" 2lthough some buildings will be serviced by indoor pico-
cells, it is expected that this site density will not provide suBcient
capacity for dense urban areas" 2 general procedure to increase
capacity would be#
6eploy micro-cells using the same carrier as the macro-cell
layer where possible to accommodate hotspots"
$mplement inter-carrier hand over and move micro-cells to
separate carrier"
6eploy further micro-cells until a micro-cell layer has been
established in the dense urban area"
$dentify micro-cells that could bene1cially re-use the macro-
cell carrier"
The micro-cells that can re-use the macro-cell carrier bene1cially
are those that enjoy geographic isolation from the macro-cell sites
as was the case in deploying micro-cells in a single carrier network"
Aowever, it is important to note that the conceptually simple
process of deploying a micro-cell layer at a separate carrier
frequency results in a great capacity jump as the capacity of the
micro-cell layer in both uplink and downlink directions is
signi1cantly greater than that for the macro-cell layer" $ndividual
cells will have a greater capacity +due to higher loading factor on
the uplink and higher orthogonality and lower interference on the
downlink, and the cell density will be typically double that of the
macro-cell layer" Thus, implementing a separate micro-cell layer
will lead to increasing the capacity of the network by a factor of
approximately !" 2dditional features such as the optimal re-use of
the macro-cell carrier frequency can lead to a further increase of
perhaps C<=" $t is important that the macro-cell carrier is not
simply re-used at all micro-cells" The resulting mutual interference
can lead to a reduction in network capacity if it is not undertaken
carefully"
/ierarchical ell Structures /ierarchical ell Structures
C The su"scri"er density that can "e
accommodated "y a macro/cell layer will have a
limit 6,00 3rlangs);m
2
7!
C .igher densities is usually due to DhotspotsF that
can "e treated separately'
C O$$ice "uildings' in/"uilding solution reuired
C 8oot"all stadiums% railway stations etc!' micro/cell
can "e deployed!
Multicarrier deployment
/ierarchical ell Structures /ierarchical ell Structures
C 1trategy
C &eploy micro/cells using the same carrier as the
macro/cell layer where possi"le to accommodate
hotspots!
C Implement inter/carrier hand over and move micro/
cells to separate carrier!
C &eploy $urther micro/cells until a micro/cell layer has
"een esta"lished in the dense ur"an area!
C Identi$y micro/cells that could "ene$icially re/use the
macro/cell carrier!
Multicarrier deployment
/ierarchical ell Structures /ierarchical ell Structures
C Micro/cells re/using the macro/cell $reuency!
C Micro/cells that are suita"ly physically isolated $rom
the macro/cell layer can "ene$icially re/use the
macro/cell $reuency!
C This should "e done only i$ the micro/cell using its
own $reuency "ecomes overloaded!
C &eploying an unnecessary e9tra $reuency will
H Reduce power availa"le to highly/used carrier
H Introduce e9tra downlin; inter$erence due to pilot and
common channels!
Multicarrier deployment
apacit( "ncreases apacit( "ncreases
C Macro/cell layer' ,00 3rlangs >T per ;m
2
6single
carrier7!
C Macro/cell layer plus continuous micro/cell layer
at separate $reuency' 300 3rlangs >T per ;m
2
!
C Macro/cell layer% micro/cell layer at separate
$reuency plus re/use o$ macro/cell $reuency
at selected cells' 3*0 3rlangs >T per ;m
2
!
C In/"uilding solutions will provide additional
capacity!
Multicarrier deployment
,5555 /and over 0etween carriers
The process of hand over between carriers is similar in principle to
that of hand over to KM except that#
$t is a &multi-service'I for example LT calls can hand over"
$t is &two-way'I an active call can hand over in both directions"
The main indicator is pilot quality, with 8c5$o levels triggering,
1rstly, compressed mode operation +8vent Dd, and then +if the
other carrier pilot is of acceptable quality, hand over +8vent !a,"
"nter "nter- -+reBuenc( hand over +reBuenc( hand over
C .and over is DhardF!
C =ased on 3c)Io comparison
C 3vent 2d' enter compressed mode!
C 3vent 3a' attempt hard hand over!
C Can "e Dtwo/wayF on an active call!
Multicarrier deployment
!0"mplementin' 3iversit(
S(stems
!05! "ntroduction
6iversity is a well-established method of improving the quality of a
communication channel" $t traditionally means employing more
than one receive antenna and then combining the signal
+sometimes merely selecting the one with the larger amplitude, so
that the outcome is superior to that which would be obtained
without diversity" (ombining has usually taken place at 0-" $n
UMT networks receive diversity actually employs multiple
receivers allowing the signals to be combined at base band" This
gives an improvement in the value of 8
b
53
@
which, in turn gives an
improvement in both coverage and capacity"
2nother innovative feature of UMT networks is the ability to utilise
transmit diversity" This is not so e%ective as receive diversity but,
nevertheless, can provide 8
b
53
@
improvements of greater than C d9
+compared to > d9 improvements possible for uplink diversity,"
!05$ 3e+inition o+ %adin'
%adin' %adin'
C 3lectromagnetic signals will interact% causing addition and
su"traction o$ their $ield strengths
C 8ast $ading signal strength changes are due to relative
motion and local scattering o"Gects such as "uildings%
$oliage% etc! and change rapidly over short distances!
C Typically Multipath inter$erence results $rom $ast $ading
C 8ading o$ the signal $ollows a Rayleigh distri"ution
C 1low $ading is the change in the local mean signal strength
as larger distances are covered!
C 8ading o$ the signal is a log/normal distri"ution
C The resultant signal at the #ode = and @3 antenna will "e
su"Gect to rapid and deep $ading
#iversity
3iversit( 3iversit(
C 1ignals $rom multiple antennas 6spatial diversity7% can "e used to
reduce the e$$ects o$ $ast $ading and improve received signal
strength!
C Three common com"ining schemes used $or Rayleigh $ading
channels 68ast $ading7 are
C 1election diversity
C chooses the strongest signal power%
C 3ual gain
C com"ines the co/phased signal voltages with eual weights%
C Ma9imal ratio com"ining
C weights the co/phased signal voltages relative to their signal to
noise ratio!
#iversity
!05) *eceive 3iversit(
*eceive 3iversit( *eceive 3iversit(
C =asic idea is that% i$ two or more independent samples o$ a signal
are ta;en% these samples $ade in an uncorrelated manner!
C 3ach path can then "e thought o$ as separate and wor;ed on in isolation
C Increases the signal to inter$erence ratio% 1IR
C Allows a system to reduce the target uplin; 3")#o o$ a channel
C 1aves @3 V #ode = (ower
C 1tandard con$iguration $or :C&MA may "e two/"ranch R9 diversity
C @sing a single cross polar antenna or two vertically polarised ones!
C 1eparation o$ the vertically polarised antenna is typically a $ew
wavelengths
#iversity
cm to separation
cm
f
c
f c
40 30
16
10 2
10 3
;
8


Uplink *eceive Space 3iversit( Uplink *eceive Space 3iversit(
C 3ven i$ signal is highly correlated% coherent com"ination should yield
a"out 3 d= improvement!
C In practice a gain o$ 5 d= or more is e9pected $rom antennas
C Typical dimension ,!*m
Receive
antenna ,
Receive
antenna 2
#iversity
UMTS .dvanced ell Plannin' and Optimisation !)5
."*OM "nternational &td $00)
Uplink *eceive Space 3iversit( Uplink *eceive Space 3iversit(
C This is not DconventionalF space diversity!
C 3ach antenna is connected to a separate $inger o$ the Ra;e
receiver!
C This is possi"le due to the synchronisation and channel estimation
derived $rom the (ilot "its on the &(CC. channel!
C 3")#o is improved% rather than simply an e$$ective power gain!
#iversity
!05- Transmit 3iversit(
3ownlink Transmit 3iversit( 3ownlink Transmit 3iversit(
C @MT1 e9plicitly allows the use o$ transmit diversity $rom the "ase station
C .owever it is not possi"le to simply transmit simultaneously $rom two close antennas as
this would cause an inter$erence pattern
C Mo"ile terminals must have the capa"ility o$ implementing downlin; transmit diversity !
Transmit
antenna ,
Transmit
antenna 2
#iversity
3ownlink Transmit 3iversit( 3ownlink Transmit 3iversit(
C @MT1 8&& mode does not allow $or an accurate measure o$ the
downlin; channel using uplin; estimations
C The @3 can measure the downlin; channel and return estimates to
the #ode = H closed loop
C The alternative is coding the downlin; to allow $or the @3 to correlate
the two signals H open loop
C The (/C(IC. is transmitted $rom each antenna di$$erently
C Orthogonal signals
C Antenna , W 0%0%0%0%0%0%0%0%0% XX Y normal operation
C Antenna 2 W 0%0%,%,%,%,%0%0%0%0%,%,%,%,%0%0%0%0%,%,%
#iversity
UMTS .dvanced ell Plannin' and Optimisation !)#
."*OM "nternational &td $00)
3ownlink Transmit 3iversit( 3ownlink Transmit 3iversit(
Transmit 3iversit(
Method
3escription
Open Loop T1T& Time 1witched Transmit antenna
&iversity $or 1C. only
Open Loop 1TT& 1pace Time "loc; coding Transmit
antenna &iversity
Closed Loop Mode , &i$$erent Orthogonal (ilots
C(IC. 4 1/C(IC.
Closed Loop Mode 2 1ame (ilot

C The $ollowing methods are suggested in the @MT1
standards to avoid the pro"lem o$ the inter$erence
#iversity
Time Switched Transmit 3iversit( 7TST38 +or S/ Time Switched Transmit 3iversit( 7TST38 +or S/
C 3ven num"ered slots transmitted on Antenna ,% odd
num"ered slots on Antenna 2
Antenna ,
Antenna 2
(/1C.
1lot Z0 1lot Z,
1lot Z,5
1lot Z2
(/1C.
(/1C. (/1C.
1/1C.
1/1C.
1/1C. 1/1C.
#iversity
Space Time Transmit 3iversit( 7STT38 Space Time Transmit 3iversit( 7STT38
C 1TT& encoding is optional in @TRA#! 1TT&
support is mandatory at the @3
C Channel coding% rate matching and
interleaving is done as in the non/diversity
mode!
C 1TT& encoding is applied on "loc;s o$ 5
consecutive channel "its
C h is the impulse channel response o$ each
antenna
#iversity
<
2 1 1
<
2 2
<
2 2 1
<
1 1
2 2
<
1 1
<
2 2
1 2 2 1 1 1
= = =
= = =
) (
) (
r h r h S
r h r h S
n h S h S r T t r
n h S h S r t r

+
+ + +
+ +
.nal(sis o+ STT3 .nal(sis o+ STT3
C 1TT& encoding e$$ectively spreads a data "it across more than one "it period!
C This leads to a general improvement in noise per$ormance!
C 8urther% it allows a stronger signal $rom one antenna to dominate!
"0 ", "2 "3
/"2 "3 "0 /",
"0/"2 ",4"3 "04"2 "3/",
(rocessing alternate "its will e9tract the data
#iversity
Antenna ,
Antenna 2
Com"ination
UMTS .dvanced ell Plannin' and Optimisation !),
."*OM "nternational &td $00)
.nal(sis o+ STT3 .nal(sis o+ STT3
C The 1pace/time com"ining generates sym"ols that are proportional to the
sum o$ the powers $rom "oth antennas
#iversity
losed &oop Mode losed &oop Mode
C Channel coding% interleaving and spreading are done as
in non/diversity mode
C The spread comple9 valued signal is $ed to "oth T[
antenna "ranches% and weighted with antenna speci$ic
weight $actors w, and w2!
C The weight $actors are determined "y the @3% and
signalled using the 8=I $ield o$ uplin; &(CC.
6&edicated (hysical Control Channel7!
1 radio *ra!e> 5 ? 10 !s
Pilot
N
pilot
bits
5PC
N
5PC
bits
-lot @0 -lot @1 -lot @i -lot @14
5slot ? 26:0 c'ips( 10 bits
*
9PCC1
AB
N
AB
bits
5ACB
N
5ACB
bits
#iversity
losed &oop Mode losed &oop Mode
1pread)scram"le
"
1
"
2
&(C.
&(CC.
&(&C.
R9
R9

C(IC.
,
T9

C(IC.
2
Ant
,
Ant
2
T9
:eight -eneration
"
1
"
2
9eter!ine AB !essage
*ro!8plin& 9PCC1
#iversity
losed &oop Mode losed &oop Mode
#iversity
C Closed Loop mode ,
C The phase o$ one antenna is adGusted relative to the other
C @sing , "it accuracy per slot
C 8eed"ac; rate is ,*00 .A
C Closed Loop mode 2
C Relative phase adGusted using 3 "it accuracy
C Amplitude adGusted using , "it
C 8eed"ac; rate is ,*00 .A
UMTS .dvanced ell Plannin' and Optimisation !-!
."*OM "nternational &td $00)
3ownlink 3ownlink E0 E09No reduction 9No reduction
#iversity
3!* d= 0!0 d= ,!0 d= ,!* d= Closed Loop ,
3!0 d= 0!* d= 0!* d= ,!0 d= Open Loop
(edestrian A
XX 3;m)h
XXXXXX
X!!,20 ;m)h
>ehicular A
X!!*0;m)h
Modi$ied
3;m)hXX!
&iversity
Mode
1ource Radio #etwor; (lanning and Optimisation $or @MT1% \aana Laiho et al
C 1lower speeds and lower multipath inter$erence produce the "est results
Transmit 3iversit( Transmit 3iversit( - - onclusions onclusions
#iversity
C &epends on @3 per$ormance
C 3stimate o$ channel impulse and 1IR
C Main "ene$it is reduction in downlin; 3")#o
C #o advantage in pro"lematic time and multipath environments
C *0;m)h // 3")#o only 0!*d= "etter in open/loop mode
C ,20;m)h // 3")#o no real improvement
C MicrocellJs will "ene$it $rom T9&iversity
C =eam $orming pro"lems associated with location
!055 Multi-User 3etection MU3
)ne major advantage that the downlink has in a UMT network is
the use of orthogonal codes to reduce the interference e%ect of
other traBc and control channels" This relies on the fact that the
downlink channels can be easily synchronised as they originate
from the same point" The same sort of cancellation is not possible
on the uplink as the transmission delay is di%erent for each user"
MU6 helps to provide some interference cancellation by performing
an inverse transform on the message contained in interfering
channels and then removing that from the input of the wanted
signal" $t is a highly sophisticated method and its potential is yet to
be fully realised" Aowever, a C d9 improvement in uplink
performance can be recorded +which can lead to useful coverage
and capacity increases," 3ote that MU6 is only e%ective at a
serving cell, the interference e%ect on neighbouring cells is not
reduced"

Multi Multi- -User 3etection User 3etection
C Multi/@ser detection 6M@&7 is a method used to improve the
per$ormance o$ the receiver "y reducing the noise
contri"utions $rom other C&MA users!
C The concept is "ased on the $act that noise $rom C&MA users%
although usually appro9imated with A:-# characteristics%
inherently consists o$ coherent signals!
C M@& reception decodes a num"er o$ users simultaneously
and su"tracts their noise contri"utions $rom the others
C 3ssentially this results in a more sensitive receiver
#iversity
UMTS .dvanced ell Plannin' and Optimisation !-)
."*OM "nternational &td $00)
Multi Multi- -User 3etection User 3etection
C Mid ,<?0s research showed that Goint% optimal% ma9imum/li;elihood decoding o$
all users out per$ormed matched $ilter alternatives!
C The pro"lem was the e9ponential increase in processing as the num"er o$
simultaneous users went up! 6 >iter"i trellis techniues 7
C Current research interests
C 1u"optimal linear receivers
C &ata/aided minimum mean suared 6MM137 linear receivers
C =lind 6 nondata/aided 7 MM13 receiver
C #on/linear multiuser detection
C Multistage inter$erence cancellation% parallel and serial% (IC V 1IC
#iversity
Multi Multi- -User 3etection User 3etection
C >iter"i decoding uses past sym"ol ;nowledge to weight present
and $uture choices
C Multiuser decoding has the added comple9ity o$ having present
Qother userJ inter$ering sym"ols
C There$ore some decision as to the inter$ering sym"ols must "e
made
C &ue to the comple9ity% multiuser detection is more li;ely to e9ist
in the #ode =
#iversity
Multi Multi- -User 3etection User 3etection
C Multiuser detection reduces the need $or tight power control
C (ower control is still important to the per$ormance o$ the M@&
system
C =est per$ormance used with short spreading codes% repeating
every sym"ol! 6 &ownlin; 7
C Can "e used with long spreading codes% pseudorandom
seuences which are much longer than the sym"ol duration!
6@plin;7
#iversity
>isualisin' >isualisin' the Processin' Aain w9o MU3 the Processin' Aain w9o MU3
:).A :).A :).A
3
c
I
o
1ignal
Intra/cell #oise
Inter/cell #oise
Before
Spreading
After
Spreading
With Noise
$
$
$
:).A
After
Despreading
/Correlation
$
:).A
3
"
#
o
Post
Filtering
(No MUD)
$
d=:).A
3
"
#
o
3
"
)#
o
$
#iversity
UMTS .dvanced ell Plannin' and Optimisation !-5
."*OM "nternational &td $00)
>isualisin' >isualisin' the Processin' Aain with MU3 the Processin' Aain with MU3
:).A
1ignal
Inter/cell #oise
After
Despreading
/Correlation
Post
Filtering
$
Other @sers
3
"
#
o
:).A
$
3
"
#
o
:).A
$
3
"
#
o
:).A
$
3
"
#
o
:).A
$
Because of MUD the contriution of the other users to the
Noise is !educed"
#t is not co$pletel% eli$inated ecause of the inaccuracies of
the Multiple access interference esti$ation"
#iversity
!056 Predictin' the E++ect o+ 3i++erent overa'e
and apacit( Enhancement 3evices
$t is clear that adding certain devices, such as mast head ampli1ers
or diversity receivers will improve network performance" Aowever,
we need to be able to quantify any likely improvement in order to
undertake a cost-bene1t analysis" 2s a starting point we shall
consider an isolated cell that is serving voice users delivering a bit
rate of CDD@@ bps at an 8
b
53
@
of > d9 on the uplink and the
downlink" .ith an uplink 3oise 0ise of ! d9 the cell can
accommodate a link loss of C!! d9"
This information alone is suBcient to suggest that the pole capacity
is C<!@ kbps on the uplink and !;DD kbps on the downlink
+assuming an orthogonality value of @"?," 2n uplink 3oise 0ise of !
d9 would suggest that ?! voice users are seen as a full load for the
cell" The loading factor on the downlink would be estimated to be
only D@= suggesting a 3oise 0ise 1gure of C d9" $f !? d9m of
common channel and pilot power is transmitted, the e%ect at the
mobile receiver would be that of a -G> d9m interference power if
the mobiles are at a path loss of CD? d9" $f the noise 7oor of the
receiver is -C@C d9m then the overall &noise plus interference' level
would be -G!"D d9m" $f a 3oise 0ise of C d9 must be produced,
then an e%ective traBc channel power of -GG"D d9m +actual receive
power -G<"D d9m, must be received" This would necessitate a
transmit power of !@"; d9m if all users were at a path loss N d9 less
than the cell edge +which is de1ned by a link loss of C!! d9,"
Ouick check downlink analysis" !@"; d9m corresponds to CD"; d9m
per user +if there are ?! users," 0eceived power per user is -CC!"D
d9m" 8%ective 3oise 4ower is -GD"D d9m +given a 30 of C d9,"
Thus wideband 30 is -DC"@ d9" 4rocessing gain of D< d9 will
restore the required 8
b
53
@
value of > d9"
Aaving carried out and understood the mechanism of this
calculation it is possible to predict the e%ect of capacity
enhancement devices such as uplink diversity" .hen considering
whether or not to use such devices it is important that their
purpose is made clear" -or example, is maximising capacity or
maximising coverage range our goal +or is it a combination of the
two aims,E 2dditionally, the a%ect on the downlink must be
assessed"
(onsider, as an example, the e%ect of implementing uplink
diversity on this cell" The e%ect is to reduce the target 8
b
53
@
value
by ! d9" $f maximising capacity +whilst keeping the coverage range
1xed, is taken to be our goal then it is possible to increase the 30
limit by ! d9 to ? d9 and then note that the pole capacity on the
uplink has doubled to !@?@ kbps" The loading factor of N<= means
that a throughput of DDG@ kbps is possible, equivalent to C;; voice
users" This represents a dramatic increase on the previous value of
?D users" Aowever, there has been no help o%ered on the
downlink" The pole capacity in this direction remains unchanged at
!;DD kbps" Thus a loading factor of ?@= will be imposed causing a
3oise 0ise of > d9" The e%ective TraBc (hannel 4ower required to
cause this 3oise 0ise will be -GC"< d9m, an actual received power of
-;N"< d9m" The total traBc channel transmit power would have to
be !;"< d9m +C<"; d9m per user," This is a signi1cant increase
over the previous value of !@"; d9m" 3otice that the amount of
power required by each user has increased signi1cantly"
2lternatively, if may be that uplink diversity has been introduced
with the goal of increasing the range of the cell keeping its capacity
constant" $f that is the case the new pole capacity of !@?@ kbps can
be used to calculate a reduced loading factor of D<=, which
represents a noise rise of C"D d9" Thus the cell coverage range can
be increased by >"; d9" Thus a typical user can be thought of as
having a path loss of C!C"; d9 to the cell" The result of this is that
the interference e%ect of the pilot and common channels is
reduced" Aowever, the fact that users are at a greater distance
UMTS .dvanced ell Plannin' and Optimisation !-#
."*OM "nternational &td $00)
means that the power requirements will be greater, although not >
d9 greater" (alculations show that the TraBc (hannel power
requirement will rise from the initial value of !@"; d9m to !D"@ d9m"
$t is possible to use similar techniques to predict the e%ect of using
mast head ampli1ers and of implementing downlink diversity"
Predictin' the E++ects Predictin' the E++ects
#iversity
C It is important to "e a"le to
predict the coverage and
capacity e$$ects o$
introducing a $eature such as
uplin; diversity into a cell!
C As a starting point we will
consider an isolated cell that
is serving voice users
delivering a "itrate o$ ,2200
"ps in "oth directions at an
3")#o o$ 5 d=!
C :e shall assume that the
orthogonality $actor is 0!+!
C Ma9imum lin; loss is ta;en to
"e ,33 d= with the Daverage
userF on the downlin; having
a lin; loss o$ ,2+ d=!
C Common Channel and (ilot
(ower ta;en to "e 33 d=m
each 6total 3+ d=m7!
C Mo"ile noise $loor is /,0,
d=m!
3ownlink alculations 3ownlink alculations
#iversity
C #oise 8loor o$ Mo"ile is /,0, d=m
C Common and (ilot Channels
received at a level o$ 3+ H ,2+ K /<0
d=m!
C Orthogonality reduces this "y 5 d=
6,0log],/0!+^K/57! Thus euivalent
is /<5 d=m!
C /<5 d=m 4 6/,0, d=m7 K /<3!2 d=m
C The pole capacity o$ the &L has
"een calculated as 3?22 ;"ps!
Throughput o$ 0?* ;"ps would "e a
loading $actor o$ 20M and a #R o$ ,
d=!
C Tra$$ic channel power has to
produce this #oise Rise!
#oise plus inter$erence
K /<3!2 d=m
3ownlink alculations 3ownlink alculations
#iversity
C #oise plus inter$erence plus tra$$ic
channel power must "e /<2!2 d=m!
C 3$$ective tra$$ic channel power must
"e /<2!2 d=m H 6/<3!2 d=m7K/<<!,
d=m!
C =ut tra$$ic channel power will "ene$it
$rom orthogonality! Actual received
tra$$ic channel power must "e /<*!,
d=m!
C Transmitted tra$$ic channel power
must total /<*!,4,2+K30!< d=m
C Con$idence chec;' +3 users' ,2!?
d=m per user' R9 power per user is
/,,3!2 d=m! #oise plus
inter$erence K /<2!2 d=m! 1#R K /
2, d=! (rocessing -ain K 2* d=!
3")#o K 5 d= as reuired!
#oise plus inter$erence
plus tra$$ic channel
power K /<2!2 d=m
Actual received tra$$ic
channel power K /<*!,
d=m
Reuired transmit
tra$$ic channel power K
30!< d=m!
"ntroducin' U& 3iversit( "ntroducin' U& 3iversit(
#iversity
C #ow we will introduce @L diversity
and prioritise capacity% ;eeping the
range the same!
C @L 3")#o improvement assumed to
"e 3 d=!
C (ole capacity on @L is now 30+0
;"psL on &L it remains at 3?22 ;"ps!
C #R limit can "e increased on @L
$rom 3 d= to + d=! Throughput on
@L increased to 22<0 ;"ps 6!11
voice users7!
C Loading $actor on &L is now +0M' a
#R o$ 5 d=!
C 3$$ective Tra$$ic Channel power is
now reuired to "e /?<!2 d=m H
6/<3!2 d=m7K/<,!* d=m!
C Actual Tra$$ic Channel (ower
Received K /?0!* d=m!
Actual received tra$$ic
channel power K /?0!*
d=m
Capacity on @L is
tre"led!
C Reuired Tra$$ic Channel transmit
power K 3?!* d=m 6,*!? d=m per
user7
Reuired TC. power K
3?!* d=m!
UMTS .dvanced ell Plannin' and Optimisation !-,
."*OM "nternational &td $00)
"ntroducin' U& 3iversit( "ntroducin' U& 3iversit(
#iversity
C #ow we will introduce @L diversity
and prioritise range increase%
;eeping the capacity the same!
C @L 3")#o improvement assumed to
"e 3 d=!
C (ole capacity on @L is now 30+0
;"psL on &L it remains at 3?22 ;"ps!
C @L loading $actor is now 2*M
C #R limit can "e reduced on @L $rom
3 d= to ,!2 d=!
C (ath loss can "e increased "y 5!?
d= so typical user now has lin; loss
o$ ,30!? d=!
C &L inter$erence $rom pilot and
common channel K /<?!0 d=m
C Adding thermal noise gives /<?!0
d=m 4 6/,0, d=m7 K/<+!0 d=m
@L path loss increased
"y 5!? d=!
"ntroducin' U& 3iversit( "ntroducin' U& 3iversit(
#iversity
C To give , d= #R on downlin;% the
3$$ective TC. power must "e /<*!0
d=m H6/<+!0 d=m7 K /,02!0 d=m!
C Actual Received TC. power must "e
/<?!0 d=m!
C Reuired Transmit TC. power must
"e 32 d=m!
C #ote' this has risen $rom 30!< d=m!
The ,!, d= rise in power is less than
the 5!? d= rise in path loss due to
the $act that the maGority o$ Dnoise
plus inter$erenceF at the mo"ile is
pilot and common channel power
$rom the cell!
C One conclusion is that it is the
loading that most in$luences
reuirements on the downlin; power
level!
@L path loss increased
"y 5!? d=!
Actual received tra$$ic
channel power K /<?!0
d=m
Reuired TC. power K
32!0 d=m!
"ntroducin' M/. "ntroducin' M/.
#iversity
@L #R increased "y 2
d=!
Capacity increased "y
30M
Reuired TC. power K
32!+ d=m!
C #ow we will now consider the e$$ect
o$ introducing a M.A and prioritising
capacity% ;eeping the range the
same!
C The #oise (er$ormance improvement
is assumed to "e 2 d=!
C (ole capacity on @L remains
unchanged at ,*30 ;"ps!
C #R limit can "e increased on @L $rom
3 d= to * d=! Throughput on @L
increased to ,05* ;"ps 616 voice
users7!
C Loading $actor on &L is now 20M' a
#R o$ ,!5 d=!
C 3$$ective Tra$$ic Channel power is
now reuired to "e /<,!? d=m H
6/<3!2 d=m7K/<0!5 d=m!
C Actual Tra$$ic Channel (ower
Received K /<3!5 d=m!
C Reuired Tra$$ic Channel transmit
power K 32!+ d=m 6,3!3 d=m per
user7
"ntroducin' M/. "ntroducin' M/. 2 2 prioritise prioritise covera'e covera'e
#iversity
Ma9 (L increased "y 2
d=
Capacity stays the
same
Reuired TC. power K
3,!3 d=m!
C #ow we will now consider the e$$ect
o$ introducing a M.A and prioritising
coverage% ;eeping the capacity the
same!
C The #oise (er$ormance improvement
is assumed to "e 2 d=!
C (ole capacity on @L remains
unchanged at ,*30 ;"ps!
C #R limit is unchanged' ma9imum lin;
loss now increased "y 2 d= to ,3*
d=!
C Loading $actor on &L is unchanged!
C 3$$ective Tra$$ic Channel power is
now reuired to "e /<3!? d=m H
6/<5!? d=m7K/,00!0 d=m!
C Actual Tra$$ic Channel (ower
Received K /<+!0 d=m! C Reuired Tra$$ic Channel transmit
power K 3,!3 d=m 6,3!3 d=m per
user7
UMTS .dvanced ell Plannin' and Optimisation !5!
."*OM "nternational &td $00)
!05# Multiple-0eam antennas
The fact that the target 8b53o is likely to be higher on the downlink
than on the uplink +typical values# voice, < d9 U:, N d9 6:I ?> kbps
cs data, ! d9 U:, N d9 6:, together with the fact that most common
enhancement devices, such as MA2s and diversity +and soft hand
over gain,, bene1t the U: more than the 6: suggests that the
downlink will become the limiting factor in a UMT network,
particularly as internet services are likely to generate more o%ered
traBc in the downlink direction than in the uplink" 2 more
sophisticated antenna that produces multiple beams is seen as a
powerful aid of the downlink"
Multi Multi- -0eam antennas 0eam antennas
#iversity
C The downlin; direction is li;ely to
"e the limiting $actor in air/inter$ace
capacity considerations!
C This is largely due to the higher
3")#o reuirement in this direction
owing to the lac; o$ diversity
receiver at the @3!
C >oice' @L 3")#o * d=L &L 0 d=
C >T' @L 3")#o 3 d=L &L 0 d=
C Multi/"eam antennas o$$er the
most power$ul method o$
increasing capacity in the downlin;
direction!
C They are also "ene$icial in the
uplin; direction!
Multi Multi- -0eam antennas 0eam antennas
#iversity
C A typical antenna $or use in a 3/
sector site will have a "eamwidth
o$ appro9imately 00 degrees!
C A multi/"eam antenna will
typically have 5 "eams with 20
degrees!
C In its simplest $orm it can "e
thought o$ as consisting o$ $our
separate antennas% each with a
narrow "eam 6this is unli;ely to
"e the actual con$iguration7!
C The power $or each user can "e
directed to the "est antenna!
!05#5! :eam +ormin' principles
(onsider two antenna elements placed side by side and fed from a
common source" uppose that the elements are fed in phase" The
electric 1eld contributions from each element along a line
perpendicular to a line joining the antennas will add in phase to
produce a maximum" Aowever, if we move away from the line the
distance to one element is di%erent from the distance to the other
and the electric 1elds will not add in phase" 2t a point where this
difference is half a wavelength +R5D, the two contributions will be in
antiphase and will cancel each other, producing a null"
2t the 1rst null,
2
sin

d
" o, if 6 d , the 1rst null occurs where
sinS T @"C, an angle of <"N degrees" Thus the beam width between
the two nulls on either side of the main lobe would be CC"> U"
.here the path length di%erence was R the two signals would add
in phase again, producing a peak" There would be nulls at dsinS T
@"< R, C"<R, D"<R etc" 2nd peaks where dsinSTR, DR, !R etc""
$f each antenna has a directional pattern itself, then the radiation
pattern of the antenna is modi1ed by the &array factor'"
$f instead of feeding the two elements in phase, a delay is placed in
one of the feeders, this has the e%ect of moving the main beam"
The peak direction will be where the delay in the feeder is
compensated for by an short distance from the antenna" -or
example, if the delay represents a phase-shift of G@U, the peak will
UMTS .dvanced ell Plannin' and Optimisation !5)
."*OM "nternational &td $00)
occur where the path length di%erence is R5>" 9y adjusting the
phase di%erence it is possible to steer the beam"

2ntenna arrays commonly consist of more than two elements" This
will lead to a narrower main beam for a given spacing" -urther, the
more antennas, the higher the gain of the array" This leads to the
possibility of reducing the gain of the individual elements" This
would usually be achieved by increasing the vertical beam width,
thus the antennas vertical height could be substantially reduced +in
the case of a > element array, it could be reduced to a quarter of its
height," )ne antenna array that shows promise for use in UMT
cells is a four-element array +or four separate four-element arrays,"
The typical arrangement is of an element spacing of approximately
@"?R" There would be four di%erent phase di%erences between
successive elements# VC!<U, V><U,-><U,-C!<U" These form the
&9utler matrix' such that each of the beams is orthogonal to the
other three" That is, where one has a peak, the other three have a
null" This leads to minimising interference between the beams"
This is the &four 1xed-beam' arrangement that is adopted in some
UMT networks"
Multi Multi- -0eam antennas 0eam antennas
#iversity
C A typical antenna $or use in a 3/
sector site will have a "eamwidth
o$ appro9imately 00 degrees!
C A multi/"eam antenna will
typically have 5 "eams with 20
degrees!
C In its simplest $orm it can "e
thought o$ as consisting o$ $our
separate antennas% each with a
narrow "eam 6this is unli;ely to
"e the actual con$iguration7!
C The power $or each user can "e
directed to the "est antenna!
Multi Multi- -0eam antennas 0eam antennas
#iversity
C The multi/"eam antenna will act
in a DsmartF manner with the @3
controlling the power weighting
$or its tra$$ic channel!
C In order $or the @3 to identi$y the
appropriate weighting% a di$$erent
secondary scram"ling code is
added to each "eam!
C The result is e$$ectively a uniue%
optimised "eam $or each user!
C On the uplin;% the Ra;e receiver
automatically adGusts the
weightings to their optimum
value!
(ilot,
(ilot2
(ilot3
(ilot5
:eam :eam- -+ormin' Principles +ormin' Principles
#iversity
C In practice% we do not use $our
separate 20 degree antennas
6they would "e physically uite
large7!
C Instead a single unit comprising a
$our/element array o$ 00 degree
antennas is used!
C To understand "eam/$orming
principles% it is "est to start with
two elements!
C I$ the two elements shown are $ed
in phase% there will "e a null
wherever
B
d
,,,,
2
C
(
2
6
(
2
3
(
2
sin

d
UMTS .dvanced ell Plannin' and Optimisation !55
."*OM "nternational &td $00)
:eam :eam- -+ormin' Principles +ormin' Principles
#iversity
C I$ instead o$ $eeding the elements
in phase% we introduce a delay in
one o$ the $eeders% the direction
o$ the main "eam will not "e
perpendicular to the line Goining
the antennas!
C The "eam can "e steered "y
adGusting the delay!
B
:eam :eam- -+ormin' Principles +ormin' Principles
#iversity
C A $our/element array will produce
a narrower main "eam $or a given
element spacing and also more
nulls!
C A particular group o$ $our element
arrays can $orm Dorthogonal
"eamsF!
C A set o$ orthogonal "eams have
nulls where the others have
pea;s!
C The set $orm a D=utler matri9F!
C O$$sets in the $eeders correspond
to phase shi$ts o$ _5*` and
_,3*`!
%our-element arra(
/30
/20
/,0
0
/?0 /+0 /50 /20 0 20 50 +0 ?0
.n'le 73e'rees8
*
e
l
a
t
i
v
e

A
a
i
n

7
d
:
8
The +our The +our- -element arra( element arra(
#iversity
C A multi/"eam antenna
can "e produced $rom
a single array with $our
di$$erent $eed
arrangements! This is
much more compact
than having $our
separate narrow/"eam
antennas!
*adiation Patterns +or di++erent phase
shi+ts
/3*
/30
/2*
/20
/,*
/,0
/*
0
/?0 /+0 /50 /20 0 20 50 +0 ?0
.n'le 7de'rees8
*
e
l
a
t
i
v
e

A
a
i
n

7
d
:
8
5*
/5*
/,3*
,3*
!05#5$ "mplementation in a UMTS network
$ts implementation is deceptively simple yet e%ective" )n the
uplink the combination of the signal from each U8 is achieved by a
0ake receiver with four 1ngers +eight if cross-polar diversity is
used," $n this way, the optimum weighting is applied to each of the
elements" The situation on the downlink is not as straightforward
as the diversity is at the transmitting end" The U8 has to identify
the relative strength of the signals from the individual elements"
This is achieved by allocating a di%erent secondary scrambling code
+each cell has C primary scrambling codes and C< secondary codes
available, to each of the elements in order that the U8 can
di%erentiate between them" This information is relayed back to the
cell on the uplink control channel and the cell can then adjust the
weightings of the transmit power so that the most power is
delivered to the element that provides the strongest signal"
!05#5) "mprovement +rom use o+ multiple-0eam antennas5
The improvement in air-interface performance is environment-
dependent but it is the most signi1cant method of improving down
link performance" 2s the down link is likely to become the major
factor limiting network capacity, it represents a very powerful
solution" )n the uplink, the gain provided if cross polar diversity is
included +requiring eight 02*8 1ngers in the receiver, is typically <
d9 using a single D-branch receiver +single antenna with cross-polar
UMTS .dvanced ell Plannin' and Optimisation !5#
."*OM "nternational &td $00)
diversity, as a reference" The downlink gain is between > and < d9,
depending on the characteristics of the propagation path" .hen
compared with the C to C"< d9 gain achieved by two-antenna
downlink diversity, it is clear that beam-forming represents a
hugely signi1cant capacity enhancement technique"
The typical hori/ontal spacing of C@ cm and the reduction in vertical
height possible without reducing coverage means that the four
elements can be housed in a single radome and form a physically
compact module" Aowever, it must be remembered that the
vertical beamwidth of the shorter antenna will be greater than for
an antenna of standard height"
Estimates o+ "mprovement Estimates o+ "mprovement
#iversity
C A multi/"eam antenna helps to isolate individual @3s $rom
inter$erence $rom other users!
C This increases capacity!
C On the downlin; an improvement o$ "etween 5 d= and * d= is
e9pected! This suggests an increase in the air/inter$ace
capacity o$ "etween ,+0M and 220M!
C On the uplin; an improvement o$ * d= compared with a cross/
polar diversity receiver can "e e9pected% providing an increase
o$ 220M in the air/inter$ace capacity!
C Multi/"eam antennas provide "y $ar the most e$$ective method
o$ increasing the capacity o$ a cell% particularly in the downlin;
direction!
Ph(sical "mplementation Ph(sical "mplementation
#iversity
C The vertical separation o$ the antennas is ,0 cm at $reuencies
o$ 2 -.A!
C The $our/element array should "e contained in a unit
appro9imately *0 cm in width!
C The height o$ the unit will depend on reuirements $or the
vertical "eamwidth!
C #ote the @3s are reuired "y the speci$ications to "e capa"le
o$ implementing the techniues descri"ed!
C They monitor the di$$erent secondary scram"ling codes and $eed
"ac; in$ormation on the D8=IF "its on the uplin; control channel!
!051 Smart 70eam-+ormin'8 antennas
The antenna described in the previous section is of the &1xed-
beam' type" The fact that antenna weighting can be varied for
each user does lead to the beam e%ectively being steered for a
user" Aowever, the phase relationship between the elements is
1xed" $t is possible to deploy &smart' antennas that produce user-
speci1c beams utilising variable phase shifts between elements"
This technology in not mature, however and requires some non-
standard functionality" -urther, it provides little improvement on
the performance of the 1xed-beam antenna" The adoption of the
1xed-beam approach is generally favoured"
UMTS .dvanced ell Plannin' and Optimisation !5,
."*OM "nternational &td $00)
!!"nte'ratin' E4tra Sites into
the Macro-ell &a(er
!!5! "ntroduction
2 requirement to improve coverage and5or capacity of the macro-
cell layer will inevitably arise at times" ometimes the only viable
solution is to introduce an extra site into this layer" $t may be at the
edge of the current coverage area or embedded within this area"
2lthough the new cell will be using the same frequency as the other
cells within the macro cell layer, care must be taken that its
introduction does not disrupt the network" This section explains
some of the issues that must be considered"
"nte'ratin' New Sites "nte'ratin' New Sites
$dditional Sites
C The con$iguration not only o$ the new site "ut also all
other sites in the region must "e considered!
C 2uite possi"le that on Dswitch onF the new site
ma;es networ; per$ormance worse than "e$ore!
C :e must uic;ly go $rom a Dnon/optimisedF to an
DoptimisedF situation!
C Care$ul use o$ a planning tool can lead to rapid
implementation o$ corrective action'
C &own/tilting antennas $rom old sites!
C Implementing new neigh"our lists!
!!5$ Plannin' the new site
.hether the new site is to be used to 1ll a coverage gap, extend
the coverage region or increase capacity, the general principles
that must be adopted are similar" 2t the moment of switch-on, the
region around the new site will move from an &optimised' situation
to &non-optimised'" 9efore the new site is made active, the
following issues must be planned carefully"
The coverage area of the new site" The new site will provide
coverage but also introduce interference" This can be
minimised by appropriate site con1guration, in particular
ensuring the antenna type and orientation are appropriate"
(areful use of the planning tool can achieve a near-optimum
con1guration as switch-on"
The coverage areas of existing sites in the region" The
previously-existing sites will interfere with the new site"
(areful use of the planning tool will ensure that the down-tilts
of antennas that will inevitably be required can be planned in
advance" The change to the down-tilts must be implemented
as soon as possible after the new site becomes active"
(hanges to the neighbour list" 2fter the planning tool has
been used to decide the con1guration of the new site and
those in its vicinity, the neighbour list can be planned, again
using the planning tool" This will lead to the creation of
neighbour lists to be allocated to the cells in the new site as
well as new neighbour lists for the cells in the region" 2gain
the changes to the neighbour list should be implemented as
soon as possible after the new site becomes active"
$t is important to realise that, until the region is optimised once
more, it is possible for the introduction of the new site to make the
network performance worse than before"
!!5) .ction a+ter activation o+ the new site
!!5)5! Makin' +urther drive-tests
-ollowing activation of the new site, it is important that drive tests
are carried out to investigate the radio environment in the region of
the new site" The procedure adopted is similar to that for the pre-
launch optimisation phase of the network as a whole" 6rive tests
should allow the optimisation engineer to#
(on1rm that coverage meet requirements
(heck that the interference levels are within acceptable limits
2ssess the capacity of a network"
(overage and interference checks would be in accordance with
&standard procedures'" That is, incidents where too many pilots
were recorded would be dealt with by re-con1guration as necessary"
3rive Testin' the *e'ion 3rive Testin' the *e'ion
$dditional Sites
C &rive Tests can "e used to'
C Assess coverage
C Assess inter$erence
C 8ine/tune the neigh"our lists
C 3stimate improvement in capacity
C This last point is vital' i$ we have not signi$icantly
increased the capacity o$ the networ; then we have
wasted our money!
UMTS .dvanced ell Plannin' and Optimisation !6)
."*OM "nternational &td $00)
Estimatin' apacit( "mprovement Estimatin' apacit( "mprovement
$dditional Sites
C Areas in the vicinity o$ the new site should show
improvements in 3c)Io and pilot strength
measurements!
C These can "e used to estimate a capacity
improvement!
!!5)5$ .ssessin' network capacit(
The capacity assessment involves an interpretation of the pilot and
8c5$o measurements made before and after the introduction of the
new site" The value of 8c5$o should show a de1nite improvement in
the area close to the new site" This can be interpreted as a
capacity improvement"
-or example, suppose that in a sample of locations the following
8c5$o measurements were made before and after deployment of the
new site" Typical measurements made on a quiet network are
shown below"
:ocation 0ef 8c5$o previously 8c5$o new
2 -G d9 -< d9
9 -; d9 -? d9
( -N d9 -N d9
6 -< d9 -? d9
3otice that the 8c5$o will not improve in all locations" The presence
of the new site will inevitably cause some interference within the
coverage area of adjacent cells" This means that the throughput
possible from a cell will actually drop at some areas" $t is important
that these areas are not the areas where high subscriber densities
are expected" $n order to perform a quantitative prediction on
capacity, the pilot $0 needs to be estimated at each location" This
will typically be C J > d9 better than the value of 8c5$o" The $0 of a
D watt bearer +assuming the pilot power is D watts, would be the
same as that of the pilot" $f the 8b53o of the service is known, the
bit rate possible for a D watt bearer can be calculated" $t is then a
simple extension to determine the full cell capacity if of the cell if
all C? watts available was used for traBc" The absolute values +D
watts and C? watts in this case, are not highly signi1cant" $t is the
ratio of the two powers +;#C or G d9, that a%ects the capacity in an
interference-limited environment"
2s an example, consider the 8c5$o in an unloaded cell in an area
where there is no signi1cant out-of-cell interference" $f the common
channel power is the same as the pilot power then#
d 3
0

I
E
c
4ilot $0 is calculated by separating the pilot power from the
common channel power and then considering the e%ect of
orthogonality in reducing the e%ective interference caused by
common channels" $f the common channel power equals the pilot
channel power then#
( ) + 1 log 10 3 -BR Pilot
0
I
E
c
d9, where D is the orthogonality factor"
This is the situation where there is no interference" Lalues of 8c5$o
worse than -! d9 indicate that out-of-cell power is being received at
that location" The ratio, D, of out-of-cell power to in-cell power is
given by
1 10
10
3
0

,
_

I E
c
D
" 3ote that the constant +-! in this case, in this
equation is related to the interference-free value of 8c5$o" $t must
be remembered that this value is appropriate only if the network is
unloaded" The pilot $0 in an unloaded network +SIR
zero
, can then be
determined as

,
_

1
1
log 10
D
SIR
zero
d9" 2t full load, the value of cell power will
be approximately N d9 +a factor of <, above the unloaded level +!?
d9m to >! d9m, but the non-pilot power +which is what we are
interested in, in the own-cell increases by a factor of G +typically
from D . to C; ., which equates to G"< d9," Thus the $0 at full
load +SIR
FL
, can be determined from
( )

,
_

1 ; 6
1
log 10
D
SIR
FL
d9"
This $0 is for a D watt pilot, a D watt traBc bearer would
experience the same $0" The capacity for a D watt bearer would
depend on the 8b53o" The capacity is given by
UMTS .dvanced ell Plannin' and Optimisation !65
."*OM "nternational &td $00)
&bit/s
10
3840
Capacity
10
23
0

,
_

FL b
SIR N E
$f the full power of C? watts is allocated to traBc channels, eight
times the throughput would be possible"
&bit/s
10
30C20
Capacity
10
*)llpo"er
0

,
_

FL b
SIR N E
" 0emember that the important
parameter is the ratio of traBc channel power to pilot power" 2
value of ;#C is thought to be appropriate" $n summary, the process
involved is
Measure 8c5$o in an unloaded network
6etermine the ratio of out-of-cell power to in-cell power, 6
from
1 10
10
3
0

,
_

I E
c
D
8stimate the pilot $0 that would be experienced if the
downlink of the network was fully loaded"
( )

,
_

1 ; 6
1
log 10
D
SIR
FL
d9"
8stimate the capacity of the downlink for a service based on
the target 8b53o,
&bit/s
10
30C20
Capacity
10
*)llpo"er
0

,
_

FL b
SIR N E
"
$n this way the impact of changes to the network can be assessed
on a capacity basis"
Estimatin' apacit( "mprovement Estimatin' apacit( "mprovement
$dditional Sites
C #ot all areas will "e improved!
C 3c)Io in adGacent cells can "e made worse "y
presence o$ new site! These areas should not "e
DhotspotsF!
-? d9 -< d9 6
-N d9 -N d9 (
-? d9 -; d9 9
-< d9 -G d9 2
8c5 $o new 8c5 $o previously :ocation
0ef
Usin' Usin' Ec Ec9"o to Estimate apacit( 9"o to Estimate apacit(
$dditional Sites
C In an unloaded networ;% i$ there is no inter$erence'
C 3c)Io should eual appro9imately /3 d=!
C Assumptions are that thermal noise is not signi$icant
and common channel power euals pilot power!
Usin' Usin' Ec Ec9"o to Estimate apacit( 9"o to Estimate apacit(
$dditional Sites
C In this case'
C This is the same 1IR as a 2 : 633 d=m7 tra$$ic
channel!
( ) d 1 log 10 3 -BR Pilot
0
+
I
E
c
UMTS .dvanced ell Plannin' and Optimisation !6#
."*OM "nternational &td $00)
Usin' Usin' Ec Ec9"o to Estimate apacit( 9"o to Estimate apacit(
$dditional Sites
C I$ the networ; is $ully loaded% Io will increase "y 0
d='
( ) d 1 log 10 4 -BR Pilot
0

I
E
c
Usin' Usin' Ec Ec9"o to Estimate apacit( 9"o to Estimate apacit(
$dditional Sites
C Throughput $rom 33 d=m "earer depends on 3")#o
target! 8or a 0 d= target '
( )
&bit/s 10 3840 p)t 5'ro)g'
10
C 1 log 10 4
0

,
_

I
E
c
Usin' Usin' Ec Ec9"o to Estimate apacit( 9"o to Estimate apacit(
$dditional Sites
C Throughput $rom 33 d=m "earer depends on 3")#o
target! 8or a 0 d= target '
( )
&bit/s 10 3840 p)t 5'ro)g'
10
C 1 log 10 4
0

,
_

I
E
c
onsiderin' "nter+erence onsiderin' "nter+erence
$dditional Sites
C I$ there is e9ternal inter$erence% 3c)Io will drop!
C Out o$ cell power as a $raction o$ in/cell power'
1 10
10
3
0

,
_

I E
c
D
UMTS .dvanced ell Plannin' and Optimisation !6,
."*OM "nternational &td $00)
onsiderin' "nter+erence onsiderin' "nter+erence
$dditional Sites
C (ilot 1IR 6at Aero load7 can then "e estimated as'
d
1
1
log 10

,
_

D
SIR
zero
1 10
10
3
0

,
_

I E
c
D
onsiderin' "nter+erence onsiderin' "nter+erence
$dditional Sites
C (ilot 1IR at $ull load can then "e estimated as'
d
1
1
log 10

,
_

D
SIR
zero
( )

,
_

1 ; 6
1
log 10
D
SIR
FL
C #ote' other cell power will increase "y a $actor o$ *!
non/pilot power $rom own cell will increase "y a $actor o$ <!
onsiderin' "nter+erence onsiderin' "nter+erence
$dditional Sites
C The cell capacity i$ ,+ watts was allocated to tra$$ic
power can then "e estimated as'
&bit/s
10
3840
Capacity
10
23
0

,
_

FL b
SIR N E
&bit/s
10
30C20
Capacity
10
*)llpo"er
0

,
_

FL b
SIR N E
C The capacity $rom a 2 watt "earer can then "e
estimated as'
Summar( o+ apacit( Prediction Summar( o+ apacit( Prediction
$dditional Sites
1 10
10
3
0

,
_

I E
c
D
( )
d
1 ; 6
1
log 10

,
_

D
SIR
FL
&bit/s
10
30C20
Capacity
10
*)llpo"er
0

,
_

FL b
SIR N E
Measure 8c5 $o in an unloaded network
6etermine the ratio of out-of-cell power to in-cell power, 6 from
8stimate the pilot $0 that would be experienced if the
downlink of the network was fully loaded"
8stimate the capacity of the downlink for a service based on thetarget 8b5 3o,
"
!!5)5$5!E4ample
2s an example, if 8c5$o was measured at -N d9 and orthogonality
factor, D, is assumed to be @"? and the 8b53o of the required service
was assumed to be N d9#"
UMTS .dvanced ell Plannin' and Optimisation !#!
."*OM "nternational &td $00)
( )
&bit/s 64:
10
30C20
Capacity
d 6 , 10
: , 0 1 ; 61 , 1 6
1
log 10
61 , 1 1 10
10
6 , 10 C
*)llpo"er
4 , 0

,
_

,
_

+
FL
SIR
D

The graph shows the predicted throughput at various locations for a
target 8b53o of N d9 and assuming that all the cell power is devoted
to users at the location in question"
!!5)5$5$E++ect o+ network loadin' on cell capacit(
$t is important to remember that the network is assumed to be at
full load" $f the interfering cells are transmitting at less than full
power then a higher capacity can be achieved" 2 more general
equation for the loaded $0 +instead of full load, is
( )

,
_

1 ; 6
1
log 10
D
SIR
LOADED
where is the fractional power
loading of the interfering cells on the downlink" -or a value for of
@"?, the throughput prediction in the above case would be
increased from <>? kbit5s to N<! kbit5s"
E4ample o+ apacit( Prediction E4ample o+ apacit( Prediction
$dditional Sites
2s an example, if 8c5 $o was measured at -N d9 and
orthogonality factor, , is assumed to be @"? and the 8b5 3o of
the required service was assumed to be N d9#"
( )
&bit/s 64:
10
30C20
Capacity
d 6 , 10
: , 0 1 ; 61 , 1 6
1
log 10
61 , 1 1 10
10
6 , 10 C
*)llpo"er
4 , 0

,
_

,
_

+
FL
SIR
D
>ariation o+ apacit( with unloaded >ariation o+ apacit( with unloaded Ec Ec9"o 9"o
$dditional Sites
The graph shows how the capacity at full load varies with the
unloaded 8c5 $o values"
Throu'hput vs5 unloaded Ec9"o
0!00
*00!00
,000!00
,*00!00
2000!00
/,5 /,2 /,0 /? /+ /5 /2 0
Unloaded Ec9"o 7d:8

a
p
a
c
i
t
(

7
k
0
i
t
9
s
8
3")#o K 0 d=
UMTS .dvanced ell Plannin' and Optimisation !#)
."*OM "nternational &td $00)
The E++ect o+ Nei'h0our ell &oadin' The E++ect o+ Nei'h0our ell &oadin'
$dditional Sites
$f the neighbouring +interfering, cells are not fully loaded then
the interference will be less and the capacity will be higher"
( )

,
_

1 ; 6
1
log 10
D
SIR
LOADED
.here is the fractional loading of the interfering cells"
$f fractional loading is @"?, the previous example would give a
predicted capacity of N<! kbit5 sinstead of <>? kbit5 s"
!!5)5) "nterpretin' measurements made under unknown
loadin' conditions5
The starting point for the analysis described above is obtaining
measurements of 8c5$o on an unloaded network" $f the network is
loaded to an unknown level, the values of 8c5$o recorded will not be
of any use" $t is possible, by making certain assumptions, to use
the pilot signals measured as an indicator of the level of 8c5$o in an
unloaded situation" uppose, for example, that the following pilots
levels were measured at a particular point#
(ell 4ilot trength
(ell C -;@ d9m
(ell D -;> d9m
(ell ! -;? d9m
(ell > -GC d9m
$f the pilot powers are added together the sum is
( ) ( ) ( ) ( )
( ) d! : , CC 10 10 10 10 log 10
10 / ;1 10 / 8: 10 / 84 10 / 80
+ + +

" -rom this level
it can be deduced that, under conditions of /ero traBc load, the
downlink power received would be approximately -N>"? d9m" Thus
the best server could be estimated to have an unloaded 8c5$o of
-<"? d9 and the capacity could be calculated in the manner
described" The assumptions in this analysis are#
Thermal noise is negligible" The value of thermal noise
depends upon the U8 and thus any assumption as to its level
would be inexact" Aowever, a value of -C@@ d9m is seen as
reasonable" This can usually be deemed &negligible' if the
network power is greater than -G@ d9m" Under conditions of
heavy load, this network power would be received if the link
loss was less than approximately C!? d9" The pilot strength
would be -C@! d9" This pilot strength is seen as a sensible
minimum for there to be any meaningful coverage and
therefore the assumption that thermal noise is negligible is
valid over the coverage area"
3etwork downlink power is the only signi1cant power" This
assumes that there are no other signi1cant interference
sources" This should be the situation" $ndeed, as the
maximum variation of 8c5$o should be N d9, the measured
value of 8c5$o can be compared with the estimate for an
unloaded 8c5$o and, if the di%erence is bigger than, say, < d9
an investigation into the conditions at the location under
question can be conducted"
"nterpretin' Measurements made under "nterpretin' Measurements made under
unknown loadin' conditions unknown loadin' conditions
$dditional Sites
The analysis uses 8c5 $o results for an unloaded network" This
can be deduced from pilot measurements when the level of
network loading is unknown"
$n the above situation, the unloaded total power will be ! d9
higher than the pilot power from each cell"
-GC d9m (ell >
-;? d9m (ell !
-;> d9m (ell D
-;@ d9m (ell C
4ilot trength (ell
( ) ( ) ( ) ( )
( ) d! : , C4 10 10 10 10 log 10
10 / 88 10 / 83 10 / 81 10 / CC
+ + +

UMTS .dvanced ell Plannin' and Optimisation !#5
."*OM "nternational &td $00)
"nterpretin' Measurements made under "nterpretin' Measurements made under
unknown loadin' conditions unknown loadin' conditions
$dditional Sites
$n this case the 8c5 $o of the best pilot +pilot strength J ;@ d9m,
would be predicted to be -<"? d9"
4rediction of capacity would be undertaken as before"
The assumptions behind the deduction of unloaded 8c5 $o must
be acknowledged"
( ) ( ) ( ) ( )
( ) d! : , C4 10 10 10 10 log 10
10 / 88 10 / 83 10 / 81 10 / CC
+ + +

"nterpretin' Measurements made under "nterpretin' Measurements made under
unknown loadin' conditions unknown loadin' conditions
$dditional Sites
2ssumptions#
Thermal noise is negligible"
3etwork power at least -G@ d9m"
Kenerally an acceptable assumption" 2 pilot power
of -C@! d9m would lead to a network power of about
-G@ d9m if network was heavily loaded"
)nly power is network power"
8xternal interference can cause problems"
8c5 $o is measured on drive test" 8vents where the
measured 8c5 $o is more than < d9 greater than
predicted, unloaded 8c5 $o should be investigated"

Das könnte Ihnen auch gefallen