Fast Planning of Efficient WCDMA Radio Networks

R. Hoppe 1, G. Wlfle 2, H. Buddendick 1, and F. M. Landstorfer 1

1

Institute of Radio Frequency Technology, University of Stuttgart

Pfaffenwaldring 47, 70550 Stuttgart, Germany, http://www.ihf.uni-stuttgart.de
2
AWE Communications GmbH, Moltkestr. 28, 71116 Gaertringen, Germany
http://www.awe-communications.com

I. INTRODUCTION
The radio network planning of GSM based systems
which utilize a combined TDMA/FDMA access scheme
can be summarized as indicated in figure 1. In the first step
predictions of the path loss are evaluated in order to ensure the coverage of the specified area.
path loss prediction

New challenges in radio network planning that come

along with the introduction of 3rd generation cellular
networks emerge from the demand for various bit rate
services and the characteristics of the WCDMA technique. According to the utilized principle of spreading
the different services lead to different processing gains
and in combination with the speed of the mobile to different C/I requirements.
Coverage

Abstract - The introduction of 3 rd generation cellular

networks and their various bit rate services require new
techniques in the field of radio network planning and
dimensioning. This paper describes a simulation tool on
system level in order to support the planning process by
analyzing the performance of a given network constellation. Finally, an example scenario for a coverage and
capacity analysis is presented and simulation results
gained with the system simulator are discussed.

uplink
downlink

Capacity

Fig. 2: Mutual influence of coverage and capacity

coverage

modifications

network

capacity

traffic modelling

Fig. 1: Planning of TDMA/FDMA systems

The capacity is dimensioned in the second step by
utilizing a frequency planning based on the predicted
traffic in the area taken into account. Usually only speech
service is considered which leads to constant cell capacities given by the number of available channels. Therefore
a separate planning of coverage and capacity can be performed.

In contrast to TDMA/FDMA (i.e. GSM) based systems the interference should be taken into account already in the coverage prediction, because the sensitivities of the base stations depend on the number of users
and used bit rates in all cells [1]. Furthermore, both upand downlink have to be analyzed in view of the possibility of different loads (asymmetric).
Special emphasis has to be given to the consideration of the mutual influence of coverage and capacity (as
indicated in figure 2). While the coverage is limited by
the uplink because of the maximum available transmitting
power of the mobile, the downlink sets limitations on the
capacity due to the increasing interference.
The impact of specific WCDMA features as the fast
transmit power control (TPC), the gain due to soft and
softer handover (SHO) and the implemented Rake receiver should be considered for the performance analysis as well. Apart from the soft limiting effects typical for
a CDMA system (due to interference limitations), hard
limitations (e.g. hardware limitations) should also be
taken into account within an overall performance analysis [2,3].

coverage
+ capacity
traffic
modelling

pdf / (1/s)

4e-3

serving time
speech

3e-3
2e-3
1e-3

dynamic WCDMA
simulation

0e+0
0

path loss
prediction
modifications network

Fig. 3: Radio network planning of WCDMA systems

As coverage and capacity depend on the instantaneous traffic distribution and influence each another, a dynamic simulation is required which combines the uplink
and downlink analysis in an adequate way (as indicated in
fig. 3).
II. SYSTEM SIMULATOR
The realized simulator is an efficient tool for analyzing
the performance of a given network constellation. It can be
utilized for planning the different phases throughout a
network roll-out. While the service providers intend to
offer speech and moderate bit rate services in a first step
within urban areas, the service area will be extended continuously and additionally high bit rate services will be
provided in limited regions.
The network layout can be freely configured concerning the base station scenario, including locations, antenna
configurations and service dependent parameters. Another important input is the user distribution. Realistic
scenarios for a given area based on service dependent
probability maps (mean arrival rate per m2) are applied in
the simulations. Possible scenarios for the simulation
include rural and urban areas.
Different standard services, as e.g. 12.2 kbps voice, 64
and 144 kbps as well as 384 kbps real-time data are defined
and the ratio between the user numbers of these services
can be selected according to the expected values. In order
to enable a dynamic simulation the traffic modeling includes two statistical processes. The arrival process determines the time between connection setups and the
serving process evaluates the connection duration. For
the generation of new users a Poisson process is evaluated for each service considering the location dependent
traffic distribution (e.g. hot spots). The duration of each
connection is determined randomly according to an adjustable characteristic depending on the type of service
(as indicated in figure 4).

200

400

600

t/s

Fig. 4: PDF for the serving time of the speech service

Consequently it is possible to investigate the data
user dominated case, where the service provider handles
the speech service over the existing 2nd generation network. But also the case of the service provider without
any 2nd generation infrastructure, handling the complete
traffic mix within the new system, can be studied.
A. Initialization
At the beginning of the simulation all parameters,
e.g. base station configuration and service dependent
settings, are initialized. The set of active users (location,
service, speed) is generated in each time step of the
dynamic simulation according to the traffic modeling
mentioned above. Some preprocessing is applied to
these data including the computation of the path loss
according to the utilized propagation model and the
determination of an active server table for every mobile
station.
B. Radio wave propagation
For the simulation of the various radio links highly
advanced propagation models are applied. In large
macro-cellular areas semi-empirical models based on
terrain databases using knife-edge-diffraction are utilized.
The radio wave propagation within urban environments is characterized by a multipath situation. Dominant phenomena are diffraction on building corners,
reflection at building walls and wave guiding in street
canyons. Therefore ray-optical models based on building databases have been developed [4, 5]. Due to several years of experience these models combine short
comp utation times with the high accuracy of a deterministic approach (see fig. 5 for a ray-optical prediction
indicating the different path losses). The models consider the effect of slow fading as e.g. the shadowing by
buildings or hills. The effect of fast fading is taken into
account by a corresponding fast fading margin in order
to allow the transmit power control to compensate the
notches particularly at the cell edge.

and receiver, the antenna gains as well as the path loss.

In order to combine the uplink and downlink analyses,
the deactivated links from the uplink iteration are not
considered during the downlink process. For accelerated
convergence of both processes a memory function is
included, which neglects oscillation effects.

EIRP

Eb/N0
PN
Noise Rise /
interference

Diversity gain

Channel

CDMAsystem gain

Path loss

Antenna gain

Transmitter loss

TX

Receiver loss

Pt0

Antenna gain

Pt0,max

Receiver

Fig. 6: Power budget

Fig. 5: Path loss prediction
Due to the increasing influence of the interference on
the performance of 3G systems the usage of accurate
propagation models within the planning process for calculating the path loss predictions becomes a must.
C. Uplink and downlink analyses
In the uplink analysis a specific tolerable transmit
power is assigned to every mobile station with minimum
overall interference. As the power levels required from the
different mobiles depend on the interference and also
influence it, this process needs iteration and the possibility of reducing the number of mobile stations served. As
criterion for the termination the change of the interference
power in comparison to the thermal noise power is evaluated.
The analysis in the downlink determines the total
transmitted power of each base station. For this purpose
the contributions of all established links are summed up
taking into account the sensitivities of the mobile stations
and the corresponding path losses. An additional amount
of common channel power is considered. If no base station exceeds its maximum power level, no established link
has to be cancelled and the instantaneous C/I can be calculated for every mobile station of each cell. The transmit
powers of the base stations are adjusted for every link
according to the difference between the instantaneous C/I
and the target C/I. With the updated transmitter power
levels the procedure is repeated until each transmitter
power level has converged.
Both the analysis in up- and downlink consider a
power budget (as given in fig. 6) including the noise rise,
the diversity and spreading gain, the loss of transmitter

D. Performance parameters
After the uplink and downlink analyses a coverage
test over the whole area is performed for each service by
using an additional test mobile.
The performance parameters calculated by the simulator include best server maps, coverage per service
maps and SHO areas. Output concerning the mobile
station transmit power, base station transmit power,
throughput per cell, load per cell, blocking and SHO
probability as well as number of active base stations per
mobile are given in a statistical manner, i.e. distribution,
mean value and standard deviation. The so called cell
breathing can be demonstrated to show the complex
interaction between coverage and capacity.
III. SIMULATION EXAMPLE
In order to visualize the results when utilizing the
dynamic simulator for the performance analysis of a
given WCDMA radio network the following example in
downtown Munich is presented.
A. Scenario
A fixed number of users for each service according
to table I is uniformly distributed among an area of about
10 km2, which is covered by a radio network consisting
of 16 base stations with omni-directional antennas. Half
of the base stations are lo cated below rooftop-level
(micro-cells), while the others are installed on the rooftops (mini-cells). The most important parameters concerning the system simulation are given in table II.

These values represent the typical configuration of a

WCDMA radio network.
TABLE I:
USER DISTRIBUTION

Service
12.2 kbps
64 kbps
144 kbps
384 kbps

mean offered
user number
300
75
25
8

mean served
user number
270
48
13
2

test mobile determines the coverage degree after uplink

and downlink analyses. Generally, during these iterative
processes a specific part of the offered traffic is put to
outage depending on the individual service (as indicated
in table 1).

Nevertheless, by using the simulator the influence of

the different parameters on the performance results can be
investigated very efficiently. In order to get more generalized results 1000 steps have been investigated. The users
are updated in each step, i.e. only in the first step the user
numbers correspond to the values of table 1, while in the
following steps the user numbers are modified according
to the specified traffic parameters (see chapter 2).
TABLE II:
SIMULATION PARAMETERS

Path loss model

Ray Tracing [4]
Max. mobile power
21 / 24 dBm
(speech / data)
Max. base station power
43 dBm
Common channel power
30 dBm
Max. downlink power per link
33 / 43 dBm
(speech / data)
Max. uplink load
0.8
Activity (speech / data)
0.6 / 1.0
Downlink orthogonality
0.4
Soft handover gain
3 dB
Active server table window
6 dB
Required Eb/N0 in uplink
5 dB
Required Eb/N0 in downlink
6 dB
BS / MS noise figure
5 / 7 dB
MS fast fading margin
4 dB

Fig. 7: Best server plot of the speech service

B. Results
The best server plot of this scenario after uplink and
downlink iteration is indicated in fig. 7 (for the speech
service), where the different colors correspond to different
cells. Obviously the cell areas are strongly split-up which
is typical for such a micro-cellular scenario.
Figure 8 presents the coverage area of the speech service. The white colored areas indicate that small parts of
the considered scenario are without coverage (mostly
inside buildings because of the additional penetration
loss). Nevertheless a coverage degree of 100% does not
correspond to a situation without call drops, because the

Fig. 8: Coverage area of the speech service

The uplink iteration determines the transmit powers
of the mobile stations ensuring minimum overall interference. Fig. 9 shows the required transmit powers of the
144 kbps data service averaged over all simulation steps.
Corresponding to fig. 8 there are higher values within

buildings whereas close to the base stations only low

transmit powers are necessary.
UL MS Power

tribution was performed 1000 times, which corresponds

to different snapshots of the mobile radio network.
Concerning the soft handover about 29 % of the mobiles are in connection with more than one base station,
which results in a soft handover gain due to better reception conditions (diversity). Thereby SHO is an important characteristic of 3rd generation radio networks.
In contrast to a static approach with this dynamic
simulator the cell breathing and the close relationship
between coverage and capacity can be visualized. By
analyzing the different parameters it is possible to predict the performance of a given radio network and to
investigate adequate modifications.
IV. CONCLUSIONS

Fig. 9: Required transmit powers in the uplink

(144 kbps data service)
In the downlink iteration, which is performed directly
after the uplink iteration, the power of each base station is
assigned according to the established links. Fig. 10 shows
the CDF of the base station powers for two different cells
(BS 5 represents a micro-cell while BS 8 corresponds to a
mini-cell). The differences between the diverse base stations occur because of different cell sizes due to local
propagation conditions and limited cell ranges at the borders of the considered area. The base station powers are
well below the max. allowed power [6]. As the transmit
power of the mobile stations is limited, links with high path
losses cannot be established, which leads to the reduced
powers at the base stations (coverage is uplink limited).

The introduction of 3rd generation cellular systems

and their new services requires advanced methods concerning the radio network planning and dimensioning. In
this paper a dynamic simulator for the evaluation of
cellular radio networks according to the WCDMA standard is presented. The simulator supports the planning
process by analyzing the performance of a specified
network constellation. Therefore this simulator can be
utilized to implement 3rd generation radio networks more
efficiently. An example visualizing the complex interaction of coverage and capacity is shown.
REFERENCES
[1]
[2]

[3]

CDF
0,8

[4]

CDF

0,6
0,4
0,2

[5]

BS5
BS8

0
0

10

15

20

BS Leistung / W
BS
Power / W

Fig. 10: CDF for the powers of two base stations

In order to generate the cumulative distribution functions of the various performance parameters the user dis-

[6]

H. Holma and A. Toskala: WCDMA FOR

UMTS , Wiley & Sons, 2000.
A. Wacker, J. Laiho, K. Sipil, M. Jsberg: Static
Simulator for Studying WCDMA Radio Network
Planning Issues, Proceedings of VTC99, Houston, Texas, May 1999, pp. 2436-2440.
J. Laiho, A. Wacker, K. Sipil, K. Heiska: The
Impact of the Subscriber Profile on WCDMA Radio
Network Performance, Proceedings of VTC99,
Houston, Texas, May 1999, pp. 2490-2494.
AWE Communications: Software Tool WinProp
for the Planning of Mobile Communication Networks (incl. demo -version), 2001, http://www.awecommunications.com
G.Wlfle, R. Hoppe, and F.M. Landstorfer: A Fast
and Enhanced Ray Optical Propagation Model for
Indoor and Urban Scenarios, Based on an Intelligent Preprocessing of the Database, 10th IEEE
PIMRC 1999, Osaka, Japan, Sept. 1999, F5
S. Hmlinen, H. Holma, and K. Sipil: Advanced WCDMA Radio Network Simulator, Proceedings of VTC99, Houston, Texas, May 1999,
pp. 2490-2494.