TAMPERE UNIVERSITY OF TECHNOLOGY Institute of Communications Engineering FRANCESC BORRS TOR Impact of Antenna Beamwidth, Propagation Slope and Coverage Overlapping on Capacity in WCDMA Networks Diplomity Master of Science Thesis Subject approved by the department council on 4.06.2003 Supervisor: Prof. Jukka Lempiinen
104
Embed
Impact of Antenna Beamwidth, Propagation Slope and ... · Propagation Slope and Coverage Overlapping on ... Impact of Antenna Beamwidth, Propagation Slope and Coverage ... of the
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
TAMPERE UNIVERSITY OF TECHNOLOGY Institute of Communications Engineering
FRANCESC BORRÀS TORÀ
Impact of Antenna Beamwidth,
Propagation Slope and Coverage
Overlapping on Capacity in WCDMA
Networks Diplomityö Master of Science Thesis Subject approved by the department council on 4.06.2003 Supervisor: Prof. Jukka Lempiäinen
Preface The work for this Master´s thesis was carried out in the radio communications group, in
the Institute of Communications Engineering, at Tampere University of Technology
(TUT), in the project �Planning and Topology of 3G Networks�.
I would like to thank Professor Jukka Lempiäinen for his help and my friend Jarno
Niemelä for his patience, personal abnegation and valuable support during my work. In
addition, I would like to thank all TUT personal: Sami (systems administrator), Tarja
(department secretary) and so many others for their efficiency and exquisite manner
during this wonderful year.
Furthermore, I would like to thank my roommate at TUT, Tuomo Kuusisto, for his
pleasant company, co-operation and fruitful working environment during all these
months.
I would also like to thank my home university in Barcelona, Universitat Politècnica de
Catalunya (UPC), for all these years of learning in an extraordinary environment.
Moreover, I would like to thank my sister Neus Borràs for her dedication and valuable
contribution for the fulfilment of this work and my girlfriend Mercè Gabaldà for her
patience during my stay in Finland.
Finally, I would specially like to thank my parents for their touching love, support and
understanding during my studies, since without this nothing would have been possible.
TAMPERE UNIVERSITY OF TECHNOLOGY Degree program in Information Technology Telecommunication Engineering Borràs Torà, Francesc: Impact of Antenna Beamwidth, Propagation Slope and Coverage Overlapping on Capacity in WCDMA Networks Master of Science thesis, 104 p. Examiner: Prof. Jukka Lempiäinen Institute of Communications Engineering August 2003 Keywords: UMTS, WCDMA, radio network planning, antenna beamwidth, antenna height, cell range, sectorisation Mobile communications sector has experienced a great growth during 1990�s. Nowadays, tendency is to provide global coverage and high capacity for high speed data services in a more flexible way. Universal Mobile Telecommunications System (UMTS) has been standardized to provide high capacity and global coverage. The air interface selected for UMTS is Wideband Code Division Multiple Access (WCDMA). This solution offers to operators a big number of significant advantages over alternative technologies, including increased network capacity, longer battery life for terminals and enhanced privacy for users. However, such benefits come at the cost of additional network complexity. This is why a background in GSM deployment is no guarantee of success in UMTS. Last estimates are that operator spending on the radio network planning of UMTS system will account for more than 60% of total capital expenditure. For this reason a robust network implementation (correct choices for key parameters like antenna beamwidth, antenna height, number of sectors/site, etc.) is critical if operators want to optimize capacity levels and enable multimedia services, which are a key element of the UMTS value proposition. In this work, impact on coverage and capacity of the system for different network configurations is investigated by using Nokia Networks static radio network planning tool NetAct WCDMA Planner 4.0, which uses Monte-Carlo simulations. After completed all these simulations, results are analysed and main conclusions are explained in order to make easier future works in the same area. To conclude, point out something significant: air interface technology (WCDMA), basic algorithms and radio network planning techniques are radically different respect to GSM, therefore, operators must demonstrate technical leadership and deployment experience in all these aspects. Only then can they expect to achieve the required levels of radio performance for tomorrow´s 3G services.
TIIVISTELMÄ 4
Tiivistelmä
TAMPEREEN TEKNILLINEN YLIOPISTO Tietotekniikan koulutusohjelma Tietoliikennetekniikan laitos Borràs Torà, Francesc: Antennin keilanleveyden, etenemiskertoimen ja peiton limittäisyyden vaikutus WCDMA verkon kapasiteettiin Diplomityö, 104 s. Tarkastaja: Prof. Jukka Lempiäinen Elokuu 2003 Avainsanat: UMTS, WCDMA, radioverkkosuunnittelu, antennin keilanleveys, antennin korkeus, solun säde, sektorointi Matkaviestinjärjestelmät ovat kokeneet suuren kasvun 1990-luvun aikana. Tänä päivänä suuntaus on tuoda käyttäjien ulottuville maailmanlaajuinen peitto ja korkeat datanopeudet joustavalla tavalla. Universal Mobile Telecommunication Systems (UMTS) on standartoitu täyttämään nämä odotukset korkeasta kapasiteetista ja maailmankattavasta peitosta. UMTS:n ilmarajapinnan pääsytekniikaksi valittiin laajakaistainen koodijakotekniikka (Wideband Code Division Multipe Access, WCDMA). Tämä tekniikka tarjoaa operaattoreille lukemattomia etuja vaihtoehtoisien tekniikkojen lisäksi. Näitä ovat mm. parantunut verkon kapasiteetti, matkapuhelimen akun pidentynyt käyttöaika ja käyttäjien yksityisyyden parantuminen. Parannukset ovat kuitenkin aiheuttaneet verkon kompleksisuuden kasvun. Tämän vuoksi GSM-verkon kaltaisella radioverkkosuunnittelulla ei taata menestystä UMTS-verkkosuunnittelun saralla. Viimeisimmän arvion mukaan yli 60% operaattorien kustannuksista aiheutuu UMTS-radioverkkon implementoinnista. Tämän vuoksi joustava radioverkon implmentointi (tärkeiden verkkoparametrien määrittäminen kuten antennin keilanleveyden, antennin korkeuden, sektoreiden lukumäärän jne.) on kriittistä, mikäli operaattorit haluavat optimoida verkon kapasiteetin ja tarjota multimedia palveluita, jotka ovat UMTS verkon uusia, keskeisiä ominaisuuksia. Tässä diplomityössä on tutkittu eri verkkokonfiguraatioiden vaikutusta radioverkon peitoon ja kapasitteettiin käyttämällä Nokia Networks:in Monte-Carlo �simulaatioita hyödyntäävää radioverkkosuunnitteluohjelmaa NetAct WCDMA Planner 4.0. Tulosten analysointi ja johtopäätöset on tehty helpottamaan tulevaisuuden verkkosuunnittelua. Tärkeäksi lopputulokseksi on saatu, että ilmarajapinnan pääsytekniikka (WCDMA), perus algoritmit ja radioverkkon suunnittelutekniikka ovat erilaisia verrattuna GSM:ään, ja sen vuoksi operaattoreiden on hyödynnettävä kaikkea teknista osaamista ja kokemusta näiltä saroilta. Vasta tämän jälkeen he voivat saavuttaa vaaditut suorituskykyvaatimukset huomisen 3G-palveluille.
EXTRACTE 5
Extracte
UNIVERSITAT TECNOLÒGICA DE TAMPERE Programa de graduació en Tecnologies de la Informació Enginyeria de Telecomunicació Borràs Torà, Francesc: Impacte de l´Ample de Feix de les Antenes, de les Pèrdues de Propagació i del Solapament de la Cobertura sobre la Capacitat de Xarxes WCDMA Projecte Final de Carrera, 104 p. Examinador: Prof. Jukka Lempiäinen Institut d´Enginyeria de les Comunicacions Agost 2003 Paraules clau: UMTS, WCDMA, planificació de xarxes ràdio, ample de feix, alçada d´antena, tamany de cel.la, sectorització El sector de les comunicacions mòbils ha experimentat un gran creixement durant els anys 90. Actualment, la tendència és proporcionar cobertura global i gran capacitat d´una forma més flexible per serveis de dades d´alta velocitat. El Sistema Universal de Telecomunicacions Mòbils (UMTS) ha estat estandaritzat per proporcionar alta capacitat i cobertura global. La interfície aèria sel.leccionada per UMTS és Accés Múltiple per Divisó de Codi en Banda Ampla (WCDMA). Aquesta solució ofereix als operadors un gran nombre de significatius avantatges sobre tecnologies alternatives, incloent major capacitat de la xarxa, vida més llarga per les bateries dels terminals i augment d´intimitat pels usuaris. No obstant, aquests beneficis arriben gràcies al cost de complexitat addicional de la xarxa. Aquest és el motiu pel qual l´experiència en GSM no és garantia d´èxit en UMTS. Les últimes estimacions indiquen que la despesa d´un operador en la planificació de xarxes ràdio UMTS superarà més del 60% del capital total invertit. Per aquesta raó una implementació robusta de la xarxa (el.leccions correctes per paràmetres clau com ample de feix de les antenes, alçada de les mateixes, nombre de sectors/cel.la, etc.) és crítica si els operadors volen optimitzar els nivells de capacitat i activar serveis multimèdia, els quals són un element clau del valor de la proposició UMTS. En aquest treball s´estudia l´impacte sobre la cobertura i la capacitat del sistema per diferents configuracions de la xarxa, utilitzant l´eina estàtica de planificació de xarxes ràdio de Nokia Networks, el NetAct WCDMA Planner 4.0, el qual utilitza simulacions Monte-Carlo. Després de completar aquestes simulacions, els resultats són analitzats, explicant les principals conclusions per tal de fer més fàcils futurs treballs en la mateixa àrea. Per acabar, assenyalar quelcom significatiu: la interfície aèria (WCDMA), els algoritmes bàsics i les tècniques de planificació de xarxes ràdio són radicalment diferents respecte a GSM, per tant, els operadors han de demostrar liderat tècnic i experiència en el desplegament en tots aquests aspectes. Només llavors poden esperar aconseguir els nivells de prestació exigits pels serveis de tercera generació del futur.
Next step is to estimate the coverage probability. This means that the standard
deviation for the log-normal fading and the propagation model exponent (according to the
values given in Table 3.2) must be set.
In the indoor case, the indoor loss is from 15 to 18 dB and the standard deviation
for log-normal fading margin calculation is set to 10-12 dB. In the outdoor case, typical
standard deviation value is 7 to 10 dB.
In real WCDMA cellular networks the coverage areas of cells overlap and the
mobile station is able to connect to more than just one serving cell. If more than one cell
can be detected the location probability increases and is higher than determined for a
single isolated cell.
In macro cells the base station is usually above rooftops and it is not possible to
calculate analytically the signal strength because of very complex propagation media.
Therefore, the empirical or semi empirical propagation models should be used. Once the
maximum allowed propagation loss in a cell is known, it is easy to apply any known
propagation model for the cell range estimation. The propagation model should be chosen
so that it optimum describes the propagation conditions in the area. The restrictions of the
model are related to the distance from the base station, the base station effective antenna
height, the mobile antenna height and the frequency. One typical example for macro
cellular environment is Okumura-Hata which is a empirical propagation model.
The Okumura-Hata model is widely used for coverage prediction in macro cells.
Based on measurement data made by Okumura in Tokyo in 1968, this data set was fitted
to mathematical model by Hata in 1980. Basic model was made for urban areas but
additionally correction factors for suburban areas and rural areas, irregular terrains and
for different base station and mobile station antenna heights have been applied. This
model is not applicable when the base station antenna is below rooftops.
The Okumura-Hata model can be written as shown in Eq. 4.7 [24].
llll(Eq. 4.7)
according to the values given in Table 4.3.
4.- PLANNING OF WCDMA RADIO NETWORKS 53
Value at frequency between 150 MHz and 1 GHz
Value at frequency between 1.5 and 2 GHz
A 69.55 46.30 B 26.16 33.90
Table 4.3 Okumura-Hata model parameters as a function of frequency. where hb is the base station antenna height in meters, d is the link distance in kilometers, f
is the center frequency in MHz, C is a tunable parameter which depends on propagation
environment (44.9 as a default value but it can vary between 44 and 47) and, finally, K is
an addition correction factor due to topology or morphology which has 0 as a default
value.
Figure 4.3 shows an example of path loss as a function of distance, by using
Okumura-Hata model [24].
Figure 4.3 Path loss as a function of distance by using Okumura-Hata model.
Eq. 4.8 presents an example of Okumura-Hata path loss model for an urban macro
cell with base station antenna height of 25 meters and carrier frequency of 2000 MHz.
More about this has been studied in [32].
(Eq. 4.8)
( ) [ ]( )kmsrdBL log7.359.138 ⋅+=
4.- PLANNING OF WCDMA RADIO NETWORKS 54
Table 4.4 shows typical maximum allowed path loss of existing GSM and
WCDMA systems.
GSM 900
GSM 1800
WCDMA
speech
WCDMA
64 Kbps
WCDMA
144 Kbps
Maximum path loss [dB] 160 154 156 157 154
Table 4.4 Typical maximum allowed path loss of existing GSM and WCDMA systems.
According to the power budget calculation (see Table 4.2) maximum allowed path
loss is 153.21 dB. Slow fading margin (which is typically 10 dB) has to be considered in
the cell range calculation and it should be subtracted from the maximum allowed path
loss before using Eq. 4.8. According to this, cell range cannot be more than 1.3 kms.
After choosing the cell range the coverage area can be calculated. The coverage
area for one cell in hexagonal configuration can be calculated with [28]:
(Eq. 4.9)
where S is the coverage area, r is the maximum cell range and K is a constant, depending
on the network topology. Its value changes according to the site configuration, as shown
in Table 4.5.
SITE CONFIGURATION onmi. 2-sectored 3-sectored 6-sectored
K 2.6 1.3 1.95 2.6
Table 4.5 K-values for the site area calculation.
According to the example given in Eq. 4.8, coverage area by using 6-sectored
sites is 4.4 km2/cell (26.4 km2/site). Total coverage area when 19 base stations are placed
is 500 km2, which is approximately the land area of Tampere. Large variations of this
calculated value due to, for example, coverage overlapping should be taken into account
in real implementation (if even small amount of coverage overlapping, theoretical total
coverage area, this 500 km2, will decrease heavily). For this reason, accurate choices for
those parameters concerning to the topology of the network, like antenna beamwidth, are
key technical elements in topology planning, which is explained in detail in section 4.2.3.
2rKS ⋅=
4.- PLANNING OF WCDMA RADIO NETWORKS 55
Once the site coverage area is known the site configurations in terms of channel
elements, sectors and carriers has to be selected so that the supported traffic density can
fulfil the requirements.
Special emphasis has to be given to the consideration of mutual influence of
coverage and capacity (as indicated in Figure 4.4) [29]. As it has been said before, the
coverage is limited by the uplink because of the maximum available transmission power
of the mobile while the downlink sets limitations on the capacity due to the increasing
interference level [33-34].
Figure 4.4 Mutual influence of coverage and capacity in WCDMA networks.
For this reason, in the very beginning the operator should have knowledge and
vision of the subscriber distribution and growth since it has a direct impact on the
coverage. Finding the correct configuration for the network so that the traffic
requirements are met and the network cost minimized is not a simple task: the number of
carriers, sectoring, loading, number of users and cell range all have a great impact on the
final result.
As coverage and capacity depend on the instantaneous traffic distribution and
influence each another, a simulation combining the uplink and downlink analysis in an
adequate way is required. Figure 4.5 shows how this influence affects the network design
and how this process can be done.
4.- PLANNING OF WCDMA RADIO NETWORKS 56
Figure 4.5 Impact of coverage and capacity on WCDMA network design.
The evaluation of the optimized base station locations can be done when the
planning threshold is defined. It means that the reasonable QoS level for the different
geographical locations have to be agreed: first major national areas, cities and roads
where coverage has to exist and then subareas of them such as urban and suburban areas.
The planning threshold also concerns whether the service has to be extended inside
vehicles and buildings in different areas.
The planning threshold itself is defined as in the GSM by starting from the mobile
station sensitivity (threshold is for the DL direction) and by adding the required planning
margins to the sensitivity level. The required margins in the WCDMA system are:
• slow fading margin (shadowing)
• macro diversity or soft HO gain
• power control headroom (fast fading margin)
• body loss
• antenna orientation loss
• in-vehicle or indoor penetration loss
• interference margin
The planning threshold is calculated by adding all these components to the mobile
station sensitivity.
4.- PLANNING OF WCDMA RADIO NETWORKS 57
4.2.3.- Topology Planning
As is has been said, coverage and capacity planning of UMTS WCDMA cellular
networks cannot be separated, since they are connected to each other. Topology planning
in 3rd generation networks combines coverage and capacity planning which contains
definition of site locations and configuration together with base station antenna
configuration, since these elements influence much on the service coverage and system
capacity of the UMTS network.
Service coverage and system capacity together with sufficient QoS and
economical implementation costs are the most essential factors that determine an
operator´s site density and site configuration for a given planning area. Site densities and
configuration of a UMTS network mainly determine coverage and capacity of a site. In
urban environments, the traffic requirements are much higher than in rural areas and thus
the site density and configuration is different for these cases. Moreover, used
implementation strategy for site configuration defines the over all coverage and capacity
of that particular site. Site locations, the number of sectors and their directions together
with antenna configuration have to be considered in such a manner that given service
coverage, system capacity and QoS requirements are satisfied with reasonable
implementation costs.
Coverage and capacity planning are linked together via power budget calculation
and load equations. This phenomenon makes it impossible to handle coverage and
capacity separately. Because of coverage depends on the loading of the network, system
level simulations are needed in order to determine the performance of UMTS network in
different planning scenarios.
The capacity of UMTS network is known to be interference limited, i.e. soft
capacity limited. For these systems, the Erlang capacity cannot be directly calculated
from Erlang-B formula, since it would give too pessimistic results. It over-estimates the
capacity need since each service is handled alone in the system calculations. If the system
is code limited, the capacity can be estimated from Erlang-B model. In code limited
situation the noise rise in the network is not causing outage or blocking, since the
communication between links is limited by the number of codes. Hence, in code limited
4.- PLANNING OF WCDMA RADIO NETWORKS 58
situation the behaviour of WCDMA network is like FDMA/TDMA based systems and
therefore Erlang-B formulas can be used. In noise limited situations the communication is
limited due to noise rise of the system, not due to the available number of codes.
In contrast to GSM networks, the capacity of UMTS network is said to be soft
blocked. In soft blocked networks the interference rise in a cell causes the blocked calls
instead of the blocking would be caused by the lack of available traffic channels or in
case of UMTS available number of codes.
Main elements in UMTS topology planning are sectoring, antenna beamwidth,
site separation, antenna height and tilting.
The term sectoring refers to increasing the number of sectors belonging to a site
[35]. In existing cellular networks, 3-sectored sites are commonly used. It has been
proposed and researched in many papers that even higher sectoring order, like 6-sectored
sites, would bring coverage and capacity enhancements for UMTS networks.
Omnidirectional base station antennas are typically used in small micro cells or
in indoor cells. Two-sectored base stations are used mainly in sectored micro cells or to
provide roadside coverage. Standard macrocellular solution for low or average loaded
networks would be use of 3-sectored sites and in macrocellular environment for high
capacity needs, 6-sectored sites would provide the best solution.
Sectoring is used in UMTS system to increase the system capacity as well as the
service coverage. It seems to be intuitive that adding more antennas to base station site
configuration increases the capacity of the site. Increasing number of sectors at the base
station site requires also place for addition hardware: when the number of sectors is
doubled, the amount of hardware is also doubled.
The effect of sectoring has two perspectives: if widebeam antennas are used,
coverage threshold is better but the interference level is also higher. Three and six-
sectored sites are depicted in Figure 4.6.
4.- PLANNING OF WCDMA RADIO NETWORKS 59
Figure 4.6 Cell structure for three and six sectors/site. Base station antenna beamwidth plays an important role in UMTS network
performance. With proper antenna beamwidth, especially the number of softer HO
connections can be controlled. Figure 4.7 shows the effect of the antenna beamwidth over
sector overlapping.
Figure 4.7 Coverage vs interference level.
Overlapped areas are possible softer HO areas, which are needed in UMTS
network in order to maintain the interference level as low as possible during sector HO
procedure. But too large softer HO areas consumes limited radio resources of the base
station. Wider base station antenna beamwidth also increases the interference level of the
4.- PLANNING OF WCDMA RADIO NETWORKS 60
neighbouring sector and thus reduces the capacity. The importance of base station
antenna beamwidth is emphasized in higher sectoring order [36].
Site separation is another key element of UMTS topology planning. Having sites
close to each other (big overlap between the cell sites) means that achieved coverage is
good in indoor as well as in outdoor locations. Conversely, this means also higher
interference levels in the network and decreased capacity values [37]. The capacity also
decreases due to higher number of soft HO connections. When the base stations are
placed far apart the cell ranges belong too long and this situation yields for high
transmission power for the mobiles located near the cell edges. Thus, the amount of
coverage overlapping is always on optimization tasks and a trade of between coverage
and capacity requirements.
Base station antenna height affects on the propagation signal near the base station
antenna. Until certain distance, the propagation near the base station antenna happens
with propagation slope of 20 dB/dec. This distance where the propagation slope changes
is called breakpoint distance. After this point, the propagation slope is determined by the
environment.
Use of higher antenna positions yields for larger coverage areas but also for
higher interference levels for surrounding cells [38]. Without tilting the base station
antenna, the interference level of the neighbouring cells increases and the capacity
decreases.
Antenna tilt is a key parameter in controlling interference and it is used, either in a
mechanical way or in an electrical manner, in order to minimize the ratio of interference
[39]. Tilt also affects the throughput at a site, making it a key differentiator when it
comes to QoS. Changing the elevation pattern offers a great opportunity for optimization.
Antenna tilt has not been used in this work.
When one of these elements is changed, the performance of the system is also
changing: sector overlapping grows with antenna height and antenna beamwidth and
when higher site separation is used the interference level between cells is lower but path
loss is higher. These are only some examples about how those parameters concerning to
the topology of the network are affecting to the system performance. More about that can
be found in [40].
4.- PLANNING OF WCDMA RADIO NETWORKS 61
Used values in simulations for those key elements of the topology planning are
shown in Table 4.6. These values have been chosen because of typical macro cellular
network layout and configuration are used. As it has been said, antenna tilt has not been
used in this work.
ANTENNA
BEAMWIDTH
[degree]
SITE
SEPARATION
[kilometers]
ANTENNA
HEIGHT
[meters]
3-SECTOR CASE 65 – 90 1.5 – 2.0 – 2.5 25 – 45
6-SECTOR CASE 33 – 65 1.5 – 2.0 – 2.5 25 – 45
Table 4.6 Simulation values of the network topology parameters.
Simulations will combine all these values in two different scenarios (three and six
sectors/site) and the objective will be to study their impact on capacity in the designed
WCDMA network. .
4.3.- OPTIMIZATION
WCDMA system needs, like GSM, continuous monitoring because the mobile
users´ location and traffic behaviour varies all the time. This monitoring requirement is
only emphasized in the WCDMA because the traffic demand can vary strongly and this
variation influences directly the radio network quality. The better and more accurately the
traffic amount and locations can be modelled the better and more efficiently the radio
network can be designed and implemented.
Main indicators that should be monitored are traffic, soft HO percentage, average
transmitted and received power, drop calls, handovers per call and per cell, throughput
and BER [26].
4.4.- CELL TYPES
For optimal UMTS performance, it is proposed that UMTS network is planned by
using a hierarchical cell structure (HCS): macro, micro and pico cells [41-43]. In general,
QoS and capacity requirements need to be guaranteed in the smallest cells, which are the
most critical cells. A possible use of the hierarchical cell structure is shown in Figure 4.8.
4.- PLANNING OF WCDMA RADIO NETWORKS 62
Figure 4.8 A hierarchical cell scenario in UMTS.
Large cells guarantee a continuous coverage for fast moving mobiles while small
cells are necessary to achieve good spectrum efficiency and high capacity for hot spot
areas. With flexible deployment, it could be possible for an operator to redeploy pico cell
channels for macro cells outside of urban cells in some locations.
The FDD macro cellular network provides a wide area of coverage and it is used
for high speed movement mobiles. Micro cells are used at street level for outdoor
coverage to provide extra capacity where macro cells could not scope. Those micro cells
would not be hexagonal in shape but rather canyonlike, reflecting the topography of a
street and be typically 200 - 400 m in distance. Pico cells would be deployed mainly in
indoor areas where there is a demand for high data rate services such as laptops
networking or multimedia conferencing. Such cells may be in the order of 50 m (typical
value) in distance. A limiting factor will be the range of these terminals when used for
high data rate services. Maximum bit rate for macro cells is hoped to be 384 Kbps and
until 2 Mbps for pico cells.
Main characteristics of these cell types, which are summarized in Table 4.7, are as
follows:
4.- PLANNING OF WCDMA RADIO NETWORKS 63
MACRO CELLS
• Cell range >1 km
• High transmission powers (>10 Watt)
• High gain, directive antennas (Ga = 10 - 20 dB)
• Data rates until 384 Kbps
• Basic coverage and capacity over large area in the first phase of network roll-out
• Low isolation between cells
• High delay spread, fast moving mobiles
• Propagation phenomena very difficult to compute analytically
• Many different propagation paths
• Lots of scattering
MICRO CELLS
• Cell range 100 m to 1 km
• Medium to high transmission powers (>1Watt)
• BS cabinet can be outdoors or indoors (long cabling in some cases)
• Medium gain antennas (Ga = 5 - 10 dB, θ3dB = 60 � 120 degree)
• Data rates even higher than 384 Kbps
• Hot-spot capacity and continuous micro coverage in some cases
• Both outdoor and indoor coverage
• Good isolation between adjacent cells
� buildings isolate cells
� good spectral efficiency
• Low minimum coupling loss close to antenna (50 - 60 dB)
• Low delay spread: only few strong propagation paths
• LOS propagation: line of sight
4.- PLANNING OF WCDMA RADIO NETWORKS 64
• Street corner effect:
� when the mobile moves from
LOS to NLOS the signal strength
might drop 20 - 30 dB
� fast moving mobiles are problematic
• Very high site density up to 20 - 30 sites per km2
PICO CELLS
• Cell range from 10 m to 100 m
• Two solutions: pico BS or distributed antenna systems (DAS)
• Small (pico BS) to high (DAS) transmission powers
• Low gain antennas (Ga = 2 dB)
• Data rates until 2 Mbps
• Hot-spot capacity / indoor coverage
• Problems:
� interaction with outdoor cells: power leaking
� low isolation
� different powers
• Low minimum coupling loss (35 - 50 dB)
• Very low delay spread: only few strong propagation paths → low multi path diversity
Support traffic peaks Hyper cells > 20 kms. Rural areas Macro cells 1 km. – 20 kms. Highways and suburban areas Micro cells 100 m. – 1 km. Cities and urban areas Pico cells < 100 m. In-building (offices, hotels, etc)
Table 4.7 Ranges and applications of the different UMTS cell types.
5.- SIMULATIONS 65
5.- SIMULATIONS
In this chapter the simulator is presented, including its characteristics and how is
it working. After this, simulations results are depicted, classified by number of sectors
used in the sectoring process. Finally, the best configurations for both cases of sectoring
are compared in order to know the best shape for our design.
5.- SIMULATIONS 66
5.1.- SIMULATION SETUP
Nokia´s simulator NetAct WCDMA Planner 4.0 is used in its static simulation
type. Static simulation is a method where the performance of the network is analysed
over various instances in time or snapshots, where User Equipments (UEs) are in
statistically determined places. The ability of each terminal to make its connection to the
network is calculated through an iterative process.
Various failure mechanisms are typically considered, such us maximum mobile
transmission power, maximum Node B power reached, no available channels or low pilot
Ec/Io.
The performance of the network is then analysed from the results of the snapshots
carried out. Monte-Carlo analysis has been used in this thesis as used in WCDMA
Planner. It is a form of static simulation which requires hours of computing time.
NetAct WCDMA Planner 4.0 offers also the possibility of run dynamic
simulations. In those simulations, UEs moving through the network in successive time
steps are simulated. A mobile list is generated and solved for the first time step. The
simulation may consider time to be split into chip periods, bit periods and time steps
(SNR considered). Successive time steps are then simulated and are dependent upon the
results of the previous time slot. New mobiles are simulated coming into the network and
terminating their calls.
Main advantages and disadvantages of these WCDMA simulation methods are
shown in Table 5.1
Method Accuracy Complexity Taken time
Static calculation
Not very accurate,
particularly with
global margins
Relatively
straightforward to
use once
configured
Shortest
Static simulation
Reasonable but
does not deal with
dynamic network
performance
More difficult to
configure and gives
more complicated
results
Moderate
depending on the
number of UEs
and calls
Table 5.1 Advantages and disadvantages of WCDMA simulation methods.
5.- SIMULATIONS 67
Method Accuracy Complexity Taken time
Dynamic
simulation
Quite high
assuming no bad
assumptions are
made to speed it
up
Difficult to judge
results
Extremely long if
multiple runs are
performed for
statistical validity
Table 5.1 (cont.) Advantages and disadvantages of WCDMA simulation methods.
What is Monte-Carlo simulation? Traditionally, TDMA/FDMA network planning
used static analysis and calculated the margins for a tuned propagation model in order to
protect the system of interferences. Gains were applied to allow for the soft handover
technique. However, as the level of intra cell and inter cell interference varies between
cells, this approach gave misleading results in early networks. Thus, in CDMA networks,
coverage and cell capacity are too interrelated to be predicted accurately with static
analysis to derive margins and gains.
An alternative approach has developed based around simulating networks by
using Monte-Carlo algorithms. Existent WCDMA Planner in NetAct WCDMA Planner
4.0 uses this approach as it provides a good balance between accuracy and usability.
A large number of randomized snapshots are taken of the network performance
for different UEs or terminals over time. In these snapshots, the UEs are in statistically
determined positions and generated independently for each snapshot.
The number of terminals in an active session in a pixel is determined by using a
Poisson distribution with a mean given by the number of terminals in the traffic array.
This means that the total number of terminals in a snapshot is Poisson distributed and so
it will vary from snapshot to snapshot.
These snapshots are then used in calculations to obtain statistically valid
measurements giving an estimate of the mean network performance.
An advantage of using the static Monte-Carlo simulation approach is that it takes
less time than dynamic simulation (where you look at mobiles moving through the
network). Repeated static simulation proves its value for detailed optimization of site
configurations, problem areas and radio resource management algorithms.
5.- SIMULATIONS 68
Used simulation environment is given by a digital map of Tampere area, which is
a city similar to Tarragona in Spain (number of people, geography and other
characteristics are quite the same).
Nineteen base stations (the first two interfering tiers and the central base station)
in hexagonal grid are placed in Tampere area in order to provide UMTS speech service to
the city. Figure 5.1 shows the digital map of Tampere used in the simulations and these
nineteen base stations in the case where three sectors/site have been used. The hexagon
limits the simulation area, so there are terminals only inside it.
Figure 5.1 Digital map of the simulation area.
The simulator needs as inputs a digital map, the network layout, where are
defined the number of sectors/site and the cell range; the traffic raster, which contains the
mobile distribution in the network and many other parameters which are needed by the
Monte-Carlo simulation and, finally, the site configuration, where the antenna parameters
5.- SIMULATIONS 69
(beamwidth and antenna positions) are set. All these parameters are chosen according to
the values shown in Table 4.6, in section 4.2.3. After this, the initialization phase has
concluded.
Once this values have been read, coverage predictions should be calculated by
using the propagation model (Okumura-Hata, which is explained in section 4.2.2).
Calculus of coverage predictions requires hours of computing time and it needs all the
resources of the computer.
After this, next step is starting the Monte-Carlo simulation. All the Monte-Carlo
simulations have been run with 10.000 snapshots in order to get reliable results. The total
number of terminals (users) in the network, according to its load level, is shown in Table
5.2.
LOAD LEVEL No. OF TERMINALS
Not loaded 2000
3 - SECTOR CASE Semi-loaded 3000
Loaded 4000
Not loaded 2000
6 - SECTOR CASE Semi-loaded 4000
Loaded 6000
Table 5.2 Total number of terminals in the network according to its load level.
5.2.- RESULTS
Results are classified by the number of sectors/site used in the sectoring process.
For each scenario (three and six sectors/site), obtained results are classified by the site
separation used in the simulations.
First, results are compared in order to know the performance of the network when
antenna height and antenna beamwidth are changed. Finally, only the best configurations
are compared between them in order to analyse the behaviour of the network when
different cell spacings are used.
5.- SIMULATIONS 70
5.2.1.- Scenario 1: 3-Sector Case
3-SECTORED SITES OF 1.5 km.
2000 T 3000 T 4000 T
90 degree / 25 m
Service probability [%] 100 96.9 79.6Mean # of mobiles in soft HO [%] 19.6 19.3 20.5Mean # of mobiles in softer HO [%] 12 12 13Uplink load [%] 42 61.3 67.2Other-to-own cell interference 0.94 0.942 0.914DL TX. power [dBm] 35.2 38.6 40.4Throughput [kbps/sector] 289.6 420.9 469.1Noise rise [dB] 2.39 4.18 4.88
90 degree / 45 m
Service probability [%] 97.7 79.6 55.7Mean # of mobiles in soft HO [%] 38.1 40.6 43.3Mean # of mobiles in softer HO [%] 9.6 10.5 11.4Uplink load [%] 53.5 63.6 58.8Other-to-own cell interference 1.535 1.48 1.43DL TX. power [dBm] 37.5 40.2 41.3Throughput [kbps/sector] 336.6 415.7 394.9Noise rise [dB] 3.4 4.43 3.89
65 degree / 25 m
Service probability [%] 100 98.7 86.1Mean # of mobiles in soft HO [%] 18.9 18.5 19.1Mean # of mobiles in softer HO [%] 4.8 4.7 5Uplink load [%] 38.6 57.3 66.4Other-to-own cell interference 0.744 0.751 0.729DL TX. power [dBm] 34.6 37.8 39.7Throughput [kbps/sector] 271.9 401.4 470.5Noise rise [dB] 2.14 3.77 4.81
65 degree / 45 m
Service probability [%] 98.8 86.7 65.1Mean # of mobiles in soft HO [%] 33.5 34.9 37.6Mean # of mobiles in softer HO [%] 3.7 4 4.4Uplink load [%] 48.7 62.8 62.6Other-to-own cell interference 1.212 1.188 1.153DL TX. power [dBm] 36.5 39.6 40.9Throughput [kbps/sector] 312.7 412.6 421.8Noise rise [dB] 2.97 4.385 4.32
Figure 5.2 Results of 3-sectored sites, 1.5 km. site separation.
In this case, base stations are close to each other. For this reason the interference
level is increasing quickly if wide and/or high antennas are used (degrading too much the
service probability when the network is loaded). According to the results shown in Figure
5.2, the best antenna configuration is 65 degree antenna beamwidth and 25 meters of
antenna height. Because of the distance between base stations is small, coverage is good
and this antenna configuration supplies a good interference level.
5.- SIMULATIONS 71
Figure 5.3 shows sector throughput as a function of DL traffic power for different
antenna configurations.
250
300
350
400
450
500
34 35 36 37 38 39 40 41 42
DL. AVERAGE TRAFFIC POWER [dBm]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
90_25
90_45
65_25
65_45
Figure 5.3 Sector throughput vs downlink traffic power with 3-sectored sites, 1.5 km. site separation.
When the transmitted power exceeds 40 dBm, the network starts to be saturated in
all cases, except for the best configuration (65_25), where the saturation area begins later.
Following the same tendency, Figure 5.4 shows the service probability as a function of
sector throughput, also for the different antenna configurations.
75
80
85
90
95
100
250 300 350 400 450 500
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
90_2590_4565_2565_45
Figure 5.4 Service probability vs sector throughput with 3-sectored sites, 1.5 km. site separation.
5.- SIMULATIONS 72
3-SECTORED SITES OF 2.0 km.
2000 T 3000 T 4000 T
90 degree / 25 m
Service probability [%] 99.7 96.9 82.7Mean # of mobiles in soft HO [%] 16.1 15.8 16.2Mean # of mobiles in softer HO [%] 12.4 12.3 13.1Uplink load [%] 39.4 57.8 65.5Other-to-own cell interference 0.824 0.833 0.798DL TX. power [dBm] 34.9 38.1 39.8Throughput [kbps/sector] 281.3 409.5 471Noise rise [dB] 2.22 3.84 4.7
90 degree / 45 m
Service probability [%] 99.9 93.3 69.7Mean # of mobiles in soft HO [%] 26.6 26.8 29.1Mean # of mobiles in softer HO [%] 11.9 12.2 13.6Uplink load [%] 46 64.5 64.5Other-to-own cell interference 1.131 1.128 1.086DL TX. power [dBm] 36.1 39.6 41Throughput [kbps/sector] 309.8 435.6 445.3Noise rise [dB] 2.71 4.55 4.53
65 degree / 25 m
Service probability [%] 99.6 98.5 87.8Mean # of mobiles in soft HO [%] 16.5 16 16.2Mean # of mobiles in softer HO [%] 5.1 5 5.2Uplink load [%] 36.3 53.9 63.8Other-to-own cell interference 0.645 0.658 0.63DL TX. power [dBm] 34.3 37.4 39.2Throughput [kbps/sector] 266.7 394 469.6Noise rise [dB] 2 3.47 4.52
65 degree / 45 m
Service probability [%] 99.9 97.1 79.5Mean # of mobiles in soft HO [%] 25.5 25.2 26.4Mean # of mobiles in softer HO [%] 4.4 4.3 4.7Uplink load [%] 42.2 61.4 66.7Other-to-own cell interference 0.901 0.908 0.875DL TX. power [dBm] 35.4 38.8 40.3Throughput [kbps/sector] 289.8 421.4 465.4Noise rise [dB] 2.41 4.22 4.83
Figure 5.5 Results of 3-sectored sites, 2.0 km. site separation.
This configuration is, according to the results shown in Figure 5.5, a medium case
and use of wide and high antennas is now better supported, compared to sites of 1.5
kilometers, because the degradation of the service probability, which is caused by this
use, is not so important, except in the worst case (wide and high antennas). The best
antenna configuration is still 65 degree and 25 meters. This configuration gives better
results in both aspects: service probability and capacity, especially when the network is
loaded.
5.- SIMULATIONS 73
Figure 5.6 shows sector throughput as a function of DL traffic power for different
antenna configurations.
250
300
350
400
450
500
550
34 35 36 37 38 39 40 41 42
DL. AVERAGE TRAFFIC POWER [dBm ]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
90_25
90_45
65_25
65_45
Figure 5.6 Sector throughput vs downlink traffic power with 3-sectored sites, 2.0 km. site separation.
The same effect appears again: when the transmitted power exceeds 40 dBm, the
network starts to be saturated but this time also for the best configuration. Figure 5.7
shows the service probability, much better compared to sites of 1.5 kilometers (especially
in the worst cases: 65_45 and 90_45) as a function of sector throughput.
75
80
85
90
95
100
250 300 350 400 450 500
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
90_25
90_45
65_25
65_45
Figure 5.7 Service probability vs sector throughput with 3-sectored sites, 2.0 km. site separation.
5.- SIMULATIONS 74
3-SECTORED SITES OF 2.5 km.
2000 T 3000 T 4000 T
90 degree / 25 m
Service probability [%] 99.1 95.1 80.5Mean # of mobiles in soft HO [%] 16.6 16.7 17.5Mean # of mobiles in softer HO [%] 12.6 12.6 13.6Uplink load [%] 39.2 57 64.3Other-to-own cell interference 0.86 0.866 0.832DL TX. power [dBm] 34.7 37.9 39.7Throughput [kbps/sector] 280.9 405 464.3Noise rise [dB] 2.21 3.78 4.59
90 degree / 45 m
Service probability [%] 99.9 94.3 75.9Mean # of mobiles in soft HO [%] 21.8 22 23.2Mean # of mobiles in softer HO [%] 12.6 12.6 13.7Uplink load [%] 43.5 61.7 66.4Other-to-own cell interference 1.035 1.03 0.998DL TX. power [dBm] 35.5 38.8 40.4Throughput [kbps/sector] 297.1 421.5 460.6Noise rise [dB] 2.52 4.24 4.79
65 degree / 25 m
Service probability [%] 99.1 97.1 86.1Mean # of mobiles in soft HO [%] 17.1 16.8 17.5Mean # of mobiles in softer HO [%] 5 5 5.3Uplink load [%] 35.9 53.2 62.8Other-to-own cell interference 0.672 0.684 0.656DL TX. power [dBm] 34.2 37.2 39.2Throughput [kbps/sector] 265.7 390 464.7Noise rise [dB] 1.98 3.41 4.43
65 degree / 45 m
Service probability [%] 99.9 96.6 82.3Mean # of mobiles in soft HO [%] 22.5 22.2 23Mean # of mobiles in softer HO [%] 4.5 4.5 4.8Uplink load [%] 40.2 58.4 66Other-to-own cell interference 0.839 0.842 0.816DL TX. power [dBm] 34.9 38.2 39.9Throughput [kbps/sector] 281.2 407.5 467.1Noise rise [dB] 2.27 3.89 4.77
Figure 5.8 Results of 3-sectored sites, 2.5 km. site separation.
This is the case where has been used the maximum spacing between base stations,
2.5 kilometers. For this reason, the impact of using wide and high antennas is now
minimum, as we can see in Figure 5.8. When the site separation is bigger, the
interference level is lower and, even if the network is loaded, there is not a big
degradation of the service probability, except in the worst case, which is another time 90
degree and 45 meters. However, the best configuration, especially from the capacity point
of view, has not changed: it is still 65 degree and 25 meters. But in the aspect of service
probability this is the best configuration only when there is a big number of terminals in
the network.
5.- SIMULATIONS 75
Figure 5.9 shows sector throughput as a function of DL traffic power for the
different antenna configurations.
250
300
350
400
450
500
34 35 36 37 38 39 40 41 42
DL. AVERAGE TRAFFIC POWER [dBm ]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
90_25
90_45
65_25
65_45
Figure 5.9 Sector throughput vs downlink traffic power with 3-sectored sites, 2.5 km. site separation.
This time the network starts to be saturated before the transmitted power arrives
to 40 dBm, except for 65_25 (best case) and for 90_45 (worst case), in which the capacity
decreases very quickly after the transmitted power has exceeded the level of 40.4 dBm.
To conclude this analysis, Figure 5.10 shows the service probability, which is not very
affected for the antenna configuration, as a function of sector throughput.
75
80
85
90
95
100
250 300 350 400 450 500
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
90_2590_4565_2565_45
Figure 5.10 Service probability vs sector throughput with 3-sectored sites, 2.5 km. site separation.
5.- SIMULATIONS 76
5.2.2.- Scenario 2: 6-Sector Case
6-SECTORED SITES OF 1.5 kms.
2000 T 4000 T 6000 T
33 degree / 25 m
Service probability [%] 100 100 96.8Mean # of mobiles in soft HO [%] 24.2 23.2 23Mean # of mobiles in softer HO [%] 4.1 4 3.9Uplink load [%] 20.1 40.1 58.1Other-to-own cell interference 0.79 0.803 0.801DL TX. power [dBm] 30.7 34.9 38.2Throughput [kbps/sector] 141.9 281.9 409Noise rise [dB] 0.98 2.27 3.89
33 degree / 45 m
Service probability [%] 100 99.4 85.1Mean # of mobiles in soft HO [%] 41.8 39.9 41.9Mean # of mobiles in softer HO [%] 2.3 2.3 2.3Uplink load [%] 24.8 49.1 61.4Other-to-own cell interference 1.208 1.219 1.171DL TX. power [dBm] 31.9 36.7 39.7Throughput [kbps/sector] 167.4 329.2 427.5Noise rise [dB] 1.25 3.03 4.26
65 degree / 25 m
Service probability [%] 100 99.4 83.8Mean # of mobiles in soft HO [%] 20.5 20 21Mean # of mobiles in softer HO [%] 36.3 34.5 37.1Uplink load [%] 24.9 50 63Other-to-own cell interference 1.409 1.434 1.386DL TX. power [dBm] 32 36.8 40.1Throughput [kbps/sector] 172.2 338 437.2Noise rise [dB] 1.25 3.06 4.37
65 degree / 45 m
Service probability [%] 100 92.2 64.1Mean # of mobiles in soft HO [%] 41.5 41.3 44.2Mean # of mobiles in softer HO [%] 37.1 37.4 41.9Uplink load [%] 31.3 57.6 57.8Other-to-own cell interference 2.135 2.118 1.999DL TX. power [dBm] 33.3 38.8 40.8Throughput [kbps/sector] 205.8 379.5 408.9Noise rise [dB] 1.65 3.81 3.8
Figure 5.11 Results of 6-sectored sites, 1.5 km. site separation.
If the number of sectors/site grows (in this scenario it is six) the overlapped area
is also bigger. For this reason the number of mobiles in HO is very big when 65 degree
antennas are used. According to the results shown in Figure 5.11, the best configuration
is now 33 degree antenna beamwidth and 25 meters of antenna height and its tendency is
the same than in 3 sectors: narrow and low antennas allow getting better service
5.- SIMULATIONS 77
probability and higher capacity because the interference level is lower and the coverage is
not so bad.
Figure 5.12 shows sector throughput as a function of DL traffic power for
different antenna configurations.
130
180
230
280
330
380
430
480
530
30 32 34 36 38 40 42
DL. AVERAGE TRAFFIC POWER [dBm]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
33_25
33_45
65_25
65_45
Figure 5.12 Sector throughput vs downlink traffic power with 6-sectored sites, 1.5 km. site separation.
When the transmitted power exceeds 40 dBm, the network starts to be saturated in
all cases except for the best configuration (33_25), which seems to grow even if the
network is very high loaded. But this configuration is not immune to interference, which
degrades the service probability as we can see in Figure 5.13, where is depicted the
service probability as a function of sector throughput.
70
75
80
85
90
95
100
130 180 230 280 330 380 430 480
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
33_2533_4565_2565_45
Figure 5.13 Service probability vs sector throughput with 6-sectored sites, 1.5 km. site separation.
5.- SIMULATIONS 78
6-SECTORED SITES OF 2.0 kms.
2000 T 4000 T 6000 T
33 degree / 25 m
Service probability [%] 99.8 99.7 97.5Mean # of mobiles in soft HO [%] 20.5 19.8 19.5Mean # of mobiles in softer HO [%] 4.8 4.6 4.5Uplink load [%] 18.8 37.4 54.8Other-to-own cell interference 0.681 0.696 0.701DL TX. power [dBm] 30.4 34.5 37.6Throughput [kbps/sector] 137.8 274.3 401.1Noise rise [dB] 0.91 2.08 3.57
33 degree / 45 m
Service probability [%] 100 99.9 95.1Mean # of mobiles in soft HO [%] 31.4 30.2 29.9Mean # of mobiles in softer HO [%] 3.4 3.3 3.3Uplink load [%] 21.9 43.5 61.1Other-to-own cell interference 0.936 0.944 0.933DL TX. power [dBm] 31.3 35.7 39Throughput [kbps/sector] 152.5 302.6 430.5Noise rise [dB] 1.08 2.52 4.2
65 degree / 25 m
Service probability [%] 99.9 99.6 88.3Mean # of mobiles in soft HO [%] 17.1 16.7 16.6Mean # of mobiles in softer HO [%] 34.9 33 33.8Uplink load [%] 23.4 46.8 62.2Other-to-own cell interference 1.24 1.267 1.236DL TX. power [dBm] 31.6 36.3 39.5Throughput [kbps/sector] 166.3 327.4 437.4Noise rise [dB] 1.17 2.81 4.32
65 degree / 45 m
Service probability [%] 100 99.1 79.6Mean # of mobiles in soft HO [%] 27 26.3 27.4Mean # of mobiles in softer HO [%] 37.9 36.2 39.5Uplink load [%] 27.2 53.9 64.2Other-to-own cell interference 1.633 1.65 1.592DL TX. power [dBm] 32.5 37.6 40.5Throughput [kbps/sector] 183.9 359.6 444.2Noise rise [dB] 1.38 3.41 4.5
Figure 5.14 Results of 6-sectored sites, 2.0 km. site separation.
In this case the overlapped area is lower (consequence of using higher cell radius)
respect to the case where the spacing between base stations was 1.5 kilometers. For this
reason, when 65 degree antennas are used, the number of mobiles in HO is not as big as it
was in the previous case but it is still big, especially if we do not use narrow antennas.
The best configuration in the capacity point of view is still 33 degree and 25 meters but it
has changed in the service probability aspect: now it is 33 degree and 45 meters because
of higher cell radius means lower overlapped area and this is low interference level.
5.- SIMULATIONS 79
Figure 5.15 shows sector throughput as a function of DL traffic power for
different antenna configurations.
130
180
230
280
330
380
430
480
530
30 32 34 36 38 40 42
DL. AVERAGE TRAFFIC POWER [dBm]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
33_25
33_45
65_25
65_45
Figure 5.15 Sector throughput vs downlink traffic power with 6-sectored sites, 2.0 km. site separation.
In Figure 5.15 is again clearly displayed the saturation area of the network, but in
this case when the transmitted power exceeds 40.2 dBm. Nevertheless, when the best
configurations (33_25 and 33_45) are used, sector throughput grows with the transmitted
power even if the network is very high loaded. From the service probability point of
view, both configurations are the best when the network is loaded, especially the last one,
as shown in Figure 5.16, where is depicted the service probability as a function of sector
throughput.
75
80
85
90
95
100
130 180 230 280 330 380 430 480
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
33_2533_4565_2565_45
Figure 5.16 Service probability vs sector throughput with 6-sectored sites, 2.0 km. site separation.
5.- SIMULATIONS 80
6-SECTORED SITES OF 2.5 kms.
2000 T 4000 T 6000 T
33 degree / 25 m
Service probability [%] 99.5 99.2 96.3Mean # of mobiles in soft HO [%] 18.9 17.2 17.2Mean # of mobiles in softer HO [%] 4.9 4.7 4.6Uplink load [%] 18.1 36.1 52.7Other-to-own cell interference 0.656 0.673 0.677DL TX. power [dBm] 30 34 37Throughput [kbps/sector] 133.5 265.2 385.9Noise rise [dB] 0.88 2 3.4
33 degree / 45 m
Service probability [%] 100 99.9 96.1Mean # of mobiles in soft HO [%] 24.8 23.9 23.8Mean # of mobiles in softer HO [%] 3.8 3.7 3.7Uplink load [%] 20.2 40.3 58Other-to-own cell interference 0.824 0.837 0.835DL TX. power [dBm] 30.7 34.9 38.1Throughput [kbps/sector] 142.6 283.6 408.9Noise rise [dB] 0.99 2.29 3.88
65 degree / 25 m
Service probability [%] 99.5 98.5 85.4Mean # of mobiles in soft HO [%] 16.4 16.1 16.8Mean # of mobiles in softer HO [%] 32 30.3 31.9Uplink load [%] 23 46 60Other-to-own cell interference 1.243 1.276 1.247DL TX. power [dBm] 31.3 36 39.1Throughput [kbps/sector] 161.6 316.4 418Noise rise [dB] 1.15 2.77 4.11
65 degree / 45 m
Service probability [%] 100 98.7 81.9Mean # of mobiles in soft HO [%] 21.4 20.9 21.9Mean # of mobiles in softer HO [%] 35 33.5 35.8Uplink load [%] 25.5 50.7 62.8Other-to-own cell interference 1.474 1.499 1.456DL TX. power [dBm] 32 37 40Throughput [kbps/sector] 172.2 336.8 427.3Noise rise [dB] 1.29 3.15 4.38
Figure 5.17 Results of 6-sectored sites, 2.5 km. site separation.
Using maximum spacing between base stations, a reduction of the number of
mobiles in HO (increasing the cell radius decreases the overlapped area) is displayed in
the results. But when widebeam antennas are used, (65 degree) this number of mobiles in
HO is still more than two times bigger with respect to the case where narrowbeam
antennas (33 degree) have been used. In the capacity point of view, the best configuration
is still 33 degree and 25 meters, according to the results shown in Figure 5.17. In the
service probability aspect it is another time 33 degree and 45 meters (like when 2.0
5.- SIMULATIONS 81
kilometers site separation were used), but now the difference compared to the use of 25
meters antenna positions and 33 degree antenna beamwidth is bigger, respect to that case
of 2.0 kilometers site separation.
Figure 5.18 shows sector throughput as a function of DL traffic power for
different antenna configurations.
130
180
230
280
330
380
430
480
530
30 32 34 36 38 40 42
DL. AVERAGE TRAFFIC POWER [dBm]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
33_25
33_45
65_25
65_45
Figure 5.18 Sector throughput vs downlink traffic power with 6-sectored sites, 2.5 km. site separation.
In this case a peculiar event is depicted in Figure 5.18 because the best
configuration, 33_25, which is still giving the best results in capacity aspect, seems to be
saturated when the network load level is high. From the service probability point of view,
33_45 is the best configuration for any state of the network, as shown in Figure 5.19,
where is depicted the service probability as a function of sector throughput.
75
80
85
90
95
100
130 180 230 280 330 380 430 480
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
33_2533_4565_2565_45
Figure 5.19 Service probability vs sector throughput with 6-sectored sites, 2.5 km. site separation.
5.- SIMULATIONS 82
5.2.3.- Optimum Configurations
Results for the best configuration in 3-sector case are depicted in Figure 5.20. It is
65 degree antenna beamwidth and 25 meters of antenna height. The poorest results are
obtained with 1.5 kilometers between sites because the interference level is higher due to
they are too close to each other. When the site separation is bigger, there is a little fall in
the number of mobiles in HO, because of the smaller overlapped area. For those cases
where the site separation is higher than 1.5 kilometers, capacity level is always better,
particularly for 2.0 kilometers case, which is the best in this aspect, especially when the
network is loaded, because its saturation area begins later. For this load level, it gives
almost 4% more of capacity respect to the case of 2.5 kms (the second one in this aspect).
3-sectored sites with 65 degree antennas and base station antenna height of 25 m.
2000 T 3000 T 4000 T 5000 T 5500T
1.5 kilometers between BTS
Service probability [%] 100 98.7 86.1 70.4Mean # of mobiles in soft HO [%] 18.9 18.5 19.1 19.6Mean # of mobiles in softer HO [%] 4.8 4.7 5 5.3Uplink load [%] 38.6 57.3 66.4 67.9Other-to-own cell interference 0.744 0.751 0.729 0.707DL TX. power [dBm] 34.6 37.8 39.7 40.5Throughput [kbps/sector] 271.9 401.4 470.5 484.1Noise rise [dB] 2.14 3.77 4.81 4.98
2.0 kilometers between BTS
Service probability [%] 99.6 98.5 87.8 74.7 68.7Mean # of mobiles in soft HO [%] 16.5 16 16.2 16.2 16.1Mean # of mobiles in softer HO [%] 5.1 5 5.2 5.4 5.5Uplink load [%] 36.3 53.9 63.8 67 67.5Other-to-own cell interference 0.645 0.658 0.63 0.601 0.588DL TX. power [dBm] 34.3 37.4 39.2 40 40.3Throughput [kbps/sector] 266.7 394 469.6 499.6 503.4Noise rise [dB] 2 3.47 4.52 4.9 4.95
2.5 kilometers between BTS
Service probability [%] 99.1 97.1 86.1 64.9Mean # of mobiles in soft HO [%] 17.1 16.8 17.5 18.1Mean # of mobiles in softer HO [%] 5 5 5.3 5.7Uplink load [%] 35.9 53.2 62.8 65Other-to-own cell interference 0.672 0.684 0.656 0.61DL TX. power [dBm] 34.2 37.2 39.2 40.3Throughput [kbps/sector] 265.7 390 464.7 486.2Noise rise [dB] 1.98 3.41 4.43 4.66
Figure 5.20 Results of 3-sectored sites with 65 degree antennas and base station antenna height of 25 m.
5.- SIMULATIONS 83
Figure 5.21 shows sector throughput as a function of DL traffic power for
different site separation used.
250
300
350
400
450
500
550
34 35 36 37 38 39 40 41 42
DL. AVERAGE TRAFFIC POWER [dBm]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
1.5 kms
2.0 kms
2.5 kms
Figure 5.21 Sector throughput vs downlink traffic power for the best configuration with 3-sectored sites.
In Figure 5.22 is depicted the service probability as a function of sector
throughput. Also in this aspect, best results are obtained using 2.0 kilometers between
sites but now the worst case is not with a spacing of 1.5 kilometers.
65
70
75
80
85
90
95
100
250 300 350 400 450 500
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
1.5 kms2.0 kms2.5 kms
Figure 5.22 Service probability vs sector throughput for the best configuration with 3-sectored sites.
5.- SIMULATIONS 84
Results for the best configuration in 6-sector case are depicted in Figure 5.23.
Now it is 33 degree antenna beamwidth and 25 meters of antenna height. The poorest
results are again obtained with 1.5 kilometers between sites. They follow the same
tendency, respect to the number of mobiles in HO, capacity and service probability, than
obtained results in 3-sector case when narrow/wide and lower/higher antenna positions
were used. But there are also some significant differences: Figure 5.24 shows that, on the
contrary of the other cases, there is not saturation area when 2.0 kilometers between sites
are used. According to this figure, best cases in the capacity point of view are 2.0 and 2.5
kilometers, both, because capacity levels are the same. When the network is high loaded,
they give more than 3% more of capacity respect to the case of 1.5 kilometers (the worst
one).
6-sectored sites with 33 degree antennas and base station antenna height of 25 m.
2000 T 4000 T 6000 T 8000 T 9000T
1.5 kilometers between BTS
Service probability [%] 100 100 96.8 82.2 74.8Mean # of mobiles in soft HO [%] 24.2 23.2 23 23.6 23.6Mean # of mobiles in softer HO [%] 4.1 4 3.9 4 4.1Uplink load [%] 20.1 40.1 58.1 65 66.1Other-to-own cell interference 0.79 0.803 0.801 0.762 0.745DL TX. power [dBm] 30.7 34.9 38.2 39.8 40.2Throughput [kbps/sector] 141.9 281.9 409 464.8 475.5Noise rise [dB] 0.98 2.27 3.89 4.66 4.79
2.0 kilometers between BTS
Service probability [%] 99.8 99.7 97.5 85.3 78.9Mean # of mobiles in soft HO [%] 20.5 19.8 19.5 19.1 18.7Mean # of mobiles in softer HO [%] 4.8 4.6 4.5 4.5 4.5Uplink load [%] 18.8 37.4 54.8 63.2 65.2Other-to-own cell interference 0.681 0.696 0.701 0.67 0.651DL TX. power [dBm] 30.4 34.5 37.6 39.1 39.6Throughput [kbps/sector] 137.8 274.3 401.1 465.4 481.8Noise rise [dB] 0.91 2.08 3.57 4.47 4.7
2.5 kilometers between BTS
Service probability [%] 99.5 99.2 96.3 78.4Mean # of mobiles in soft HO [%] 18.9 17.2 17.2 17.5Mean # of mobiles in softer HO [%] 4.9 4.7 4.6 4.7Uplink load [%] 18.1 36.1 52.7 63.9Other-to-own cell interference 0.656 0.673 0.677 0.645DL TX. power [dBm] 30 34 37 39.3Throughput [kbps/sector] 133.5 265.2 385.9 472.4Noise rise [dB] 0.88 2 3.4 4.61
Figure 5.23 Results of 6-sectored sites with 33 degree antennas and base station antenna height of 25 m.
5.- SIMULATIONS 85
Figure 5.24 shows sector throughput as a function of DL traffic power for
different site separation used.
130
180
230
280
330
380
430
480
530
30 32 34 36 38 40
DL. AVERAGE TRAFFIC POWER [dBm]
SEC
TOR
TH
RO
UG
HPU
T [K
bps/
sect
or]
1.5 kms
2.0 kms
2.5 kms
Figure 5.24 Sector throughput vs downlink traffic power for the best configuration with 6-sectored sites.
Service probability as a function of sector throughput is depicted in Figure 5.25.
Also from this point of view, best results are obtained using 2.0 kilometers between sites.
In this aspect, the worst case is again the same: using a site separation of 2.5 kilometers.
75
80
85
90
95
100
130 180 230 280 330 380 430 480
SECTOR THROUGHPUT [Kbps/sector]
SER
VIC
E PR
OB
AB
ILIT
Y [%
]
1.5 kms
2.0 kms
2.5 kms
Figure 5.25 Service probability vs sector throughput for the best configuration with 6-sectored sites.
6.- CONCLUSIONS 86
6.- CONCLUSIONS
Coverage and capacity planning of UMTS WCDMA cellular networks cannot be
separated, since they are connected to each other. Furthermore, uplink and downlink
directions should be considered separately due to different (asymmetric) traffic.
Topology planning phase contains definition of key parameters like base station
antenna configuration (antenna beamwidth, antenna height or tilting), number of sectors
used in the sectoring process or description of base station locations: distance between
sites, etc. These elements have great impact on the system performance and they
influence strongly on capacity and service coverage of UMTS networks. All these
parameters should be defined in order to provide good coverage and capacity levels,
fulfiling QoSrequirements with reasonable implementation costs.
UMTS network capacity is interference limited, i.e. it is a soft capacity limited
system, because the interference rise in a cell causes the blocked calls. Interference level
is displayed in the simulation results, in section 5.2, and it decreases when the overlapped
area is smaller. If tilting is not considered (it has not been used in this work), the
capacity is higher when narrow antenna beamwidth and lower antenna positions are used.
Use of high antennas is better supported for the network when the distance between base
stations and/or the number of sectors/site are increased. Thus, when spacing between sites
is 2.5 kilometers, service probability is always better using narrow antennas with higher
antenna positions. This is due to higher antenna positions give larger coverage areas but
also higher interference level to neighbouring cells; nevertheless, when large spacing
between sites is used, the effect of better coverage has greater importance than
interference level. Antenna height affects the signal propagation near the base station and
antenna beamwidth allows to control softer HO areas, i.e. number of softer HOs, which
are needed in UMTS network in order to limit the interference level during the HO
procedure but should be taken into account that too large softer HO areas consumes
limited resources of the base station. Importance of a good choice for the base station
antenna beamwidth grows with the order of sectorisation.
In the aspect of spacing between sites, best results are obtained by using 2.0
kilometers between them, which is a middle value between 1.5 kms (better coverage but
6.- CONCLUSIONS 87
too much interference and bigger number of soft HO connections due to higher
overlapped area) and 2.5 kms (the network needs more transmitted power for the mobiles
located near the cell edges, if they are not in HO area, and also could exist coverage
holes).
In the sectoring point of view, by using a bigger number of sectors/site the
capacity of the network is higher but there is also a drop in the service probability,
although these differences are not much significant. Sectoring increases the number of
softer handover connections in the network when widebeam antennas are used,
decreasing its capacity because there is a greater consumption of the base station limited
resources. Sectoring is used in UMTS networks to increase the system capacity as well as
the service coverage. In macro cellular environment, the best solution for high capacity
requirements is provided by using six-sectored sites.
About the situation when the configuration is not the optimum, in a mixed shape
(narrowbeam and high antennas or low antenna positions combined with wide antenna
beamwidth) and 3-sectored sites, the effect of choosing widebeam antennas is better
supported for the network than the effect of choosing high antenna positions, but in 6-
sectored sites it is on the contrary: results are better by using high than widebeam
antennas. This is because with six-sectored sites overlapped area grows faster with
widebeam antennas than with higher antenna positions, increasing the interference level
of the network. With 3-sectored sites, this use of widebeam antennas is not as critical as it
was when 6 sectors/site were used because cells are bigger and therefore the overlapped
area is smaller.
To conclude, best system configurations in both scenarios (three and six
sectors/site) depending on the network load level, are shown in Table 6.1.
NOT LOADED NETWORK LOADED NETWORK
3-SECTOR CASE 65º - 2.5 kms - 25 m 65º - 2.0 kms – 25 m
6-SECTOR CASE 33º - 2.5 kms – 25 m 33º - 2.0 kms – 25 m
Table 6.1 Best system configurations in three and six sectored sites.
Using 3-sectored sites, a configuration of 65 degree antenna beamwidth, 2.0
kilometers between sites and 25 meters of antenna height gives, when the network is
6.- CONCLUSIONS 88
loaded, a minimum of 17.2 Kbps/sector more of capacity (see Figure 5.20). If
throughput/user is 15 Kbps/user, this configuration allows, with the same QoS level, 65
users more in the network respect to other configurations.
Similarly, if 6-sectored sites are used, repeating the same analysis when the
network is loaded, this situation appears again, but now with 33 degree antennas, 2.0
kilometers between sites and 25 meters of antenna height. That configuration gives a
minimum of 9.4 Kbps/sector more of capacity (see Figure 5.23), which is equivalent to
71 users more in the network compared to other configurations.
In my opinion, in this point it is necessary to study the impact of tilting on
coverage and capacity of WCDMA systems because it has not been used in this work.
Maybe it can change the tendency shown by the results of this thesis, which indicate as
the best configuration for the base station antennas (when tilting is not used) narrowbeam