ADJACENT CHANNEL INTERFERENCE
Impact on the Capacity of WCDMA/FDD Networks
Daniel Figueiredo, Pedro Matos, Nuno Cota, António Rodrigues
Instituto Superior Técnico, Technical University of Lisbon, Av. Rovisco Pais, Lisbon, Portugal
Keywords: WCDMA, adjacent channel interference (ACI), FDD, spectrum management
Abstract: The adjacent channel interference (ACI) can result in a reduced network capacity in a multioperator
WCDMA/FDD environment. This paper is devoted to the study of the ACI, using a static simulator.
Simulations were performed in order to identify particular scenarios and network compositions where ACI
plays a major role in the system capacity. On the basis of the results, the authors identify the best strategy
for frequency deployment within the available spectrum. It is demonstrated that the macro carrier should be
located in the centre of the frequency band, protected from the ACI introduced by other operators. It is, in
fact, the carrier which suffers the greatest losses caused by the increase in ACI. Furthermore, the micro
carrier should be placed as close as possible to the adjacent channel of other operators in order to maximize
system capacity.
1 INTRODUCTION
At the moment radio spectrum is becoming
increasingly more occupied, making its
management a vital tool to network planning. It is
under these circumstances that the third generation
(3G) mobile communications systems emerged, and
in particular the UMTS (Universal Mobile
Telecommunications System) in Europe. The air
interface chosen for the 3G UMTS system was
WCDMA (Wideband Code Division Multiple
Access) for the paired bands FDD (Frequency
Division Duplex). The scope of this paper is to
study the impact of frequency utilization on
WCDMA/FDD networks and develop a strategy to
optimise it.
The performance of any CDMA system is
conditioned by the interference. From the various
possible sources of interference that are present in
these systems this paper focuses on the study of
adjacent channel interference (ACI) and its effect
on the overall capacity of the system.
This study consists of five main sections.
Following a brief overview of a few features of
UMTS related with the options chosen for the
simulations, to be found in Section 2, the criteria for
the choice of the tested scenarios are explained in
Section 3. Section 4 presents the results obtained
with the simulated scenarios, and final conclusions
are drawn in Section 5.
2 INTERFERENCE ISSUES IN
UMTS/FDD NETWORKS
When defining the UMTS system, 3GPP (3
rd
Generation Partnership Project) inferred that each
radio channel has a bandwidth of 5 MHz and the
channels allocated are positioned beside each other
in uplink and downlink bands separated by 190
MHz (3GPP 25.101). Each channel has its carrier
and these are assigned to the UMTS operators in the
market.
The strategy designed to reduce the
interference, thus achieving the highest possible
capacity, consists in identifying optimal spacing
between carriers in the radio spectrum available for
use. For this purpose, a specific frequency
arrangement must be considered, as it may vary
from country to country.
The initial licensed spectrum for UMTS in
FDD mode was a band with twelve carriers, both
uplink and downlink. The case considered includes
four operators, each of which has been allocated
three carriers. Within the allocated band, it is
74
Figueiredo D., Matos P., Cota N. and Rodrigues A. (2004).
ADJACENT CHANNEL INTERFERENCE - Impact on the Capacity of WCDMA/FDD Networks.
In Proceedings of the First International Conference on E-Business and Telecommunication Networks, pages 74-80
DOI: 10.5220/0001381900740080
Copyright
c
SciTePress
possible to choose the spacing between its carriers
and the distance from adjacent operator carriers.
3GPP defines for the spacing between channels a
raster of 200 kHz (3GPP 25.104), which means that
the spacing between carriers can vary in increments
of 200 kHz around 5 MHz.
As seen in Figure 1, in order to decide which
spacing should be used, more issues must be taken
into account, to prevent the carriers from
encroaching on their neighbours. Consequently,
bearing in mind the limits typically used in
simulations, it has been chosen to vary the distances
between the values 4.6 and 5.2 MHz.
Figure 1: Spacing between carriers in a UMTS system
(adapted from (Holma and Toskala, 2002))
In a WCDMA system, developed in the above
context, the interference can stem from a large
number of sources, namely, thermal noise, traffic in
the same cell, traffic in adjacent cells and traffic
from operators using adjacent cells.
Possible ways of measuring the interference
leakage between connections operating on different
carriers must be considered. As the filter is not
perfect, when transmitting in its own channel, one
carrier will send part of its power into adjacent
channels. This effect is measured as the ACLR
(Adjacent Channel Leakage Ratio). On the other
hand, the receiver filter is unable to receive only the
desired signal alone, which is why the rejection of
the adjacent channel signal is measured as ACS
(Adjacent Channel Selectivity). Moreover, when
considering the existence of two carriers which
interfering with each other, the total interference is
given as an ACIR (Adjacent Channel Interference
Ratio) and determined by (1).
(1)
11
1
A
C
S
A
CLR
ACIR
+
=
Furthermore, this source of interference can be
seen both from the uplink and from the downlink
standpoint. Consider an uplink connection, whose
ACIR is given in (2) below. As it is quite likely that
the filter in the user equipment (UE) will be poorer
than the filter in the base station (BS), the UE
ACLR dominates in the case of uplink. In
downlink, the situation is analogous, as seen in (3),
where the UE ACS dominating on this occasion.
(2)
11
1
BSUE
UL
ACSACLR
ACIR
+
=
(3)
11
1
UEBS
DL
ACSACLR
ACIR
+
=
One of the essential parameters used in the
simulation was that the value of the filter depends
on the spacing given to the channels. 3GPP defines
the filter’s mask, while identifying minimum values
for filters at 5 and 10 MHz [3, 4]. However, in this
project more realistic values were used, which
correspond to real equipment presently available.
These values may be found in Table 1.
Table 1: Values of the filters used in BS and UE
ACLR (dB) ACS (dB) Spacing
(MHz)
UE BS UE BS
5 33 60 33
10 43 65 43
70
When simulations were run using spacing
different from 5 or 10 MHz, e.g. 4.6 MHz, a
logarithmic regression is made to convert the filter
value and obtain a valid method to compare the
results.
3 SIMULATION SCENARIOS
The choice of which scenarios to study was not as
simple as it might be assumed at first glance. One of
the goals in this paper was to find scenarios where
ACI has a major role on the network’s capacity, in
order to understand the impact of placing carriers
with different spacing. In (3GPP 25.942), the
authors give an idea of the issues to be born in mind
when choosing which scenarios to simulate.
In a preliminary stage, the search started with
the study of the representative scenarios of rural and
urban environments. When simulating two
operators, BSs working with adjacent carriers were
uniformly distributed over a map. In order to
ADJACENT CHANNEL INTERFERENCE - Impact on the Capacity of WCDMA/FDD Networks
75
simulate a worst case situation, the sites of both
operators are not co-located and the interoperator
spatial offset is equal to the cell radius (Hiltunen,
2002).
It was found that the inter-frequency
interference impact on the capacity was minimal,
when compared with the intra-frequency
interference. The reason for this result lies in the
fact that there are too many BSs from the same
carrier interfering with each other.
The next step taken was to identify scenarios
where the ACI played a significant role, at least as
important as the interference coming from the
connections working on the same carrier. Following
simple scenarios, where just a few BSs and two
carriers were taken into account, the analysis
developed to encompass broader environments
simulating urban centres with many antenna sites
and three carriers.
A simple map was used as an entrance
parameter to the simulator, with no additional
information, apart from UE and BS positions. When
placing the BSs of two different carriers, one must
decide whether they are co-located, i.e. both cells
lie on the same site, or not. In the latter situation, it
is assumed that the worst case for ACI happens, i.e.
the adjacent channel site is located at the coverage
edge of the first channel cell.
The simulator used to achieve this analysis was
static, using a Monte-Carlo evaluation method. As a
result, the users were placed randomly on the map.
Following the iterative process, only the
connections with sufficient Eb/No (or signal to
noise ratio - SNR) for the appointed service were
considered to be served by the system. This
simulator was adapted from the previous one
described in (Laiho et al., 2002) and (Wacker et al.,
2001). By examining many static situations,
referred to as snapshots, network capacity is
estimated through the average number of the served
users (Povey et al., 2003).
The bit rates tested in this study were chosen in
accordance with the services expected to be offered
by operators in the first implementation phase. In
this case, 12.2 kbps with CS (Circuit Switching), 64
kbps with CS and, finally, 128 kbps in downlink
and 64 kbps in uplink using PS (Packet Switching).
The results are presented taking into account users
accessing one of these three types of services.
The UE power classes considered for
determining the maximum output power were class
3 (24 dBm) for voice and class 4 (21 dBm) for data
services (3GPP 25.101). The BS maximum output
power used was 43 dBm.
Two types of antennas were chosen to simulate
macro and micro BS: for the macro BS, tri-
sectorized antennas with 18 dBi of gain, and for the
micro BS, omni-directional antennas with 4 dBi of
gain.
Two different propagation models were
considered to calculate the path loss according to
the characteristics of the environment (both for
outdoor propagation). For rural scenarios the COST
231 Hata model was used. The main input
parameters for the model are the UE antenna
heights, 1.5 m, and BS antenna heights, 35 m. For
the urban environment the propagation model
applied was COST 231 Walfish-Ikegami. The main
parameters used are UE antenna heights, 1.5 m, BS
antenna heights, between 10 and 25 m (depending if
they are macro or micro), street width, 20 m,
building separation, 40 m, and building height, 12
m.
4 RESULTS
In the extended study that originated this paper, a
wide range of scenarios and environments were
considered (Figueiredo and Matos, 2003). Urban,
rural and motorway environment were tested using
layers containing twenty-three macro cells placed in
the form of a grid. Furthermore, eight scenarios
with only a few antennas (five at the most) were run
to simulate specific situations using macro and
micro cells. The dense urban environment was
simulated, by using macro cells layers and micro
cells to cover identified hotspots. In this paper, only
the three most significant tests will be presented.
At the end of each simulation, the outputs were
analysed. Apart from the maps indicating the BS
and UE position, the network’s capacity (measured
in average number of served users) and the capacity
loss (when compared with no ACI), parameters like
the ratio between sources of interference were also
analysed in (Figueiredo and Matos, 2003). The
interference sources considered in the results
included the interference coming from the adjacent
channel, the interference from the same channel
from neighbouring cells and the interference from
the same cell (due to the other users connected to
the same BS).
A maximum load of 50 % was allowed in the
radio interface.
4.1 Case 1: Small scale networks with
two operators
This scenario was developed to study the impact of
a new micro BS placed by an operator to cover a
hotspot in the middle of an existing network of
macro BSs from the adjacent carrier competitor.
ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS
76
The users from both operators have been located
around the centre of the area considered. The area
simulated has a high density of active clients, as
shown in Figure 2.
The number of users presented in Figure 2
corresponds to the initial users from each operator,
and are placed on the map at the simulation start.
Figure 2: Location of UE (CS 12.2) and BS (Case 1)
Table 2 shows the average number of users
served employing different channel spacing and the
percentage of loss compared to the case where no
adjacent operator exists (without ACI) for the CS
12.2 kbps service.
It has been verified that the operator 2,
covering the area with four macro BSs, is the one
that suffers most from interference. This may be
explained by the fact that users of operator 1 (micro
BS) are closer to the antenna, which therefore
makes it more difficult for them to lose the
connection. A comparison of the results obtained
from the simulations performed with the three
services tested for operator 2, is shown in the graph
presented in Figure 3.
Table 2: Results from the simulated scenario (Case 1)
CS 12.2 kbps Capacity (average
number of users)
Capacity
Loss (%)
Operator 1
Without ACI
77 0
4.6 MHz
76.1 1.17
4.8 MHz
75.8 1.56
5.0 MHz
76.7 0.39
5.2 MHz
76.7 0.39
10.0 MHz
77 0
Operator 2
Without ACI
162.1 0
4.6 MHz
44.8 72.36
4.8 MHz
104.7 35.41
5.0 MHz
126.2 22.15
5.2 MHz
134.6 16.96
10.0 MHz
155.5 4.07
Without ACI
4.6 MHz
4.8 MHz
5.0 MHz
5.2 MHz
10 MHz
0
20
40
60
80
Capacity Loss (%)
Spacing between carriers
Scenario from case 1
CS 12.2 (Voice) CS 64 PS 64/128
Figure 3: Capacity Loss of operator 2 (Case 1)
4.2 Case 2: One layer macro and two
layers micro
In the situation shown in Figure 5 several macro
BSs were placed to form a grid and cover the area
to serve users of operator 1. Four hotspot areas
(with higher user density) from both operators 1 and
2 were placed and micro BSs located to cover them.
In this case, it has been tested an environment
where three carriers coexist and interfere with each
other. Operator 1 has one carrier for macro BSs (f1)
and another for micro BSs (f2). Operator 2 has only
one carrier for micro BSs (f3). The three channels
were allocated next to each other in the radio
spectrum as shown in Table 3.
ADJACENT CHANNEL INTERFERENCE - Impact on the Capacity of WCDMA/FDD Networks
77
Scenario from case 2
1200
1250
1300
1350
1400
1450
1500
1550
4.6 MHz 4.8 MHz 5 MHz 5.2 MHz
Spacing betw een carriers macro and micro from operator 1
Average number of served
users by operator 1
4.6 MH
z
4.8 MH
z
5 MHz
5.2 MH
z
Figure 4: Capacity of operator 1 (Case 2)
Figure 5: Location of UE (CS 12.2) and BS (Case 2)
Table 3: Frequency planning with macro (M) and micro
(m) networks (Case 2)
Operator 1 Operator 2
f1 (M) f2 (m) f3 (m)
Once again the spacing considered between
each of the three channels varied within the range
4.6 to 5.2 MHz.
The results obtained from these simulations are
given in Figure 4. The graph shows the average
number of users served by operator 1, and take into
account both the users connected to the macro (f1)
and to the micro (f2) layers. As expected, it can be
seen that the network capacity rises when the
spacing between f1 and f2 widens. As both micro
layers accommodate fewer users than the macro
layer from operator 1, it is evident that the distance
between micro layers (f2 and f3) from the different
operators does not have a great impact on the
capacity of operator 1. This fact is confirmed by the
graph, since the lines are almost overlapped.
Following the analysis of these results, it is
logically preferable to choose a wider spacing
between carriers f1 and f2, in order to achieve an
increase in the capacity of operator 1. At the same
time, it is reasonable to leave the lowest distance to
the carrier from operator 2 (f3), since the damage is
imperceptible, towards optimisation of the spectrum
allocated.
4.3 Case 3: Two layers macro and
one layer micro
As in the previous case, in Case 3, the authors
tested the impact of interference among three
carriers controlled by two operators. However, in
this case, there are two major macro BSs grids from
operators 1 and 2. The same hotspots mentioned in
Spacing between
o
p
erators 1 and 2
ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS
78
Scenario from case 3
1200
1250
1300
1350
1400
1450
1500
1550
4.6 MHz 4.8 MHz 5 MHz 5.2 MHz
Spacing betw een carriers f1 and f 2 f rom operator 1
Aaverage number of users
served by operator 1
4.6 MHz
4.8 MHz
5 MHz
5.2 MHz
Figure 6: Capacity of operator 1 (Case 3)
Figure 7: Location of UE (CS 12.2) and BS (Case 3)
Case 2 are now covered with micro BSs by operator
1 only, as seen in Figure 7. Thus, the configuration
of the radio spectrum is similar, the only difference
being that there are two channels for macro BSs and
one for micro BSs, as shown in Table 4.
Table 4: Frequency planning with macro (M) and micro
(m) networks (Case 3)
Operator 1 Operator 2
f1 (M) f2 (m) f3 (M)
The results, presented in the graph from Figure
6, were obtained by using the same procedure
followed in Case 2.
As before, the network’s capacity grows as the
distance between f1 and f2 becomes larger.
However, it can now be confirmed that the spacing
between the two adjacent carriers from different
operators (f2 and f3) has a significant impact on the
overall capacity of operator 1. This feature is due to
the fact that carrier f3, from operator 2, now
accommodates a much larger number of users in its
macro layer.
In this situation the analysis has to be
considered more carefully than in the previous case.
To achieve maximum capacity in operator 1,
apparently the best solution would be to choose the
maximum spacing between carriers f1 and f2
whilst, at the same time, also leaving the highest
distance to the adjacent operator channel (f3).
However, in doing so, one is failing to take into
account the fact that each operator has three
allocated channels. Note that, in the future, it will
be valuable to use all of them to face an anticipated
traffic increase. As a result, it would wise to choose
a configuration in which the two carriers from
operator 1 are not positioned in such a way that they
occupy the free space left by hitherto unused third
channel.
Spacing between
o
p
erato
r
s 1 and 2
ADJACENT CHANNEL INTERFERENCE - Impact on the Capacity of WCDMA/FDD Networks
79
5 CONCLUSIONS
In this paper the authors studied the impact of the
ACI on a general network’s capacity. This led to
some more useful conclusions that may be applied
when planning the launch of a WCDMA/FDD radio
networks.
When considering two wide BSs grids that lie
close to each other to cover a specific area, it was
observed that the main interference source is not the
ACI, but interference from the neighbouring BSs,
working on the same channel.
From scenarios like the one presented in Case
1, it was noted that the macro BSs are more likely
to suffer from ACI when new hotspots are covered
with micro BS by a competitor operator. This fact is
explained by the longer distance between the user
and the macro BS, as compared with the latter. As
the macro carrier may suffer a greater impact on
capacity, it should be protected and placed in the
centre channel of the allocated spectrum. This
choice is irrespective of the number or type of
carriers used, assuming that the operator launching
a service uses at least one macro carrier.
It may also be seen, from these small and
specific case scenarios like Case 1, that the use of a
4.6 MHz spacing may cause critical problems,
leading to a serious reduction in the network’s
capacity. Therefore, distances between carriers of
4.6 MHz or less should never be used. Although in
the vast majority of the situations the loss may not
be that disastrous, the possibility of having certain
areas with losses above 50 % is unsustainable to an
operator.
Upon considering an available spectrum of
three carriers, and assuming that the macro carrier is
located in the centre channel, it is intended to
decide where to place the micro channel. From Case
2, where operator 2 placed a micro carrier in the
channel adjacent to the spectrum of operator 1, it
was seen that the distance between the two channels
was almost irrelevant to the overall network’s
capacity. However, when operator 2 has a macro
carrier on the channel, adjacent to operator 1, the
latter suffers the consequences of a reduction in the
distance between different operators’ channels. On
the basis of the compromise solution of not
occupying the spectrum of the three channels
allocated using two carriers only, it may be
concluded that a spacing of 5.2 MHz between f1
and f2 and 4.8 MHz between f2 and f3 is the best
option.
ACKNOWLEDGEMENTS
We would like to thank Luís Santo and Ana Claro
for their support and useful discussions that helped
to improve this work. The authors are also grateful
to Optimus for the support given to this project.
REFERENCES
Laiho, J., Wacker, A., and Novosad, T., 2002, Radio
Network Planning and Optimisation for UMTS, John
Wiley & Sons, Sussex, England
Holma, H., and Toskala, A., 2002, WCDMA for UMTS –
2
nd
Edition, John Wiley & Sons, Sussex, England
3GPP Technical Specification 25.101 v5.5.0, UE Radio
Transmission and Reception (FDD)
3GPP Technical Specification 25.104 v5.5.0, BS Radio
Transmission and Reception (FDD)
3GPP Technical Specification 25.942 v5.1.0, Radio
Frequency (RF) System Scenarios
Hiltunen, K.,2002, Interference in WCDMA Multi-
Operator Environments, Postgraduate Course in
Radio Communications 2002-2003, Helsinki
University of Technology, Finland
Wacker, A., Laiho, J., Sipilä, K., Heiska, K., and
Heikkinen, 2001, K., NPSW – MatLab
Implementation of a Static Radio Network Planning
Tool for Wideband CDMA
Povey, G., Gatzoulis, L., Stewart, L., and Band, I., 2003,
WCDMA Inter-operator Interference and “Dead
Zones”, Elektrobit (UK) Ltd, University of Edinburgh
Figueiredo, D., and Matos, P., 2003, Analysis, Impact and
Strategy of Frequency Utilisation on WCDMA/FDD
Networks (in Portuguese), Final Graduation Thesis,
Instituto Superior Técnico, Lisboa, Portugal
ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS
80