WIRELESS COMMUNICATIONS AND MOBILE COMPUTING

Wirel. Commun. Mob. Comput. (2008)

Published online in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/wcm.672
Cost analysis and efcient radio bearer selection
for multicasting over UMTS
Antonios Alexiou
1,2
, Christos Bouras
1,2
*
,y
, Vasileios Kokkinos
1,2
and Evangelos Rekkas
1,2
1
Research Academic Computer Technology Institute, N. Kazantzaki str, GR-26500 Patras, Greece
2
Computer Engineering and Informatics Department, University of Patras, GR-26500 Patras, Greece
Summary
Along with the widespread deployment of the Third Generation (3G) cellular networks, the fast-improving
capabilities of the mobile devices, content, and service providers are increasingly interested in supporting
multicast communications over wireless networks and in particular over Universal Mobile Telecommunications
System (UMTS). To this direction, the Third Generation Partnership Project (3GPP) is currently standardizing the
Multimedia Broadcast/Multicast Service (MBMS) framework of UMTS. In this paper, we present an overview of
the MBMS multicast mode of UMTS. We analytically present the multicast mode of the MBMS and analyze its
performance in terms of packet delivery cost under various network topologies, cell types, and multicast users
distributions. Furthermore, for the evaluation of the scheme, we consider different transport channels for the
transmission of the multicast data over the UMTS Terrestrial Radio-Access Network (UTRAN) interfaces. Finally,
we propose a scheme for the efcient radio bearer selection that minimizes total packet delivery cost. Copyright #
2008 John Wiley & Sons, Ltd.
KEY WORDS: UMTS; multicast in UMTS; MBMS; power control
1. Introduction
UMTS constitutes the third generation of cellular
wireless networks which aims to provide high-speed
data access along with real time voice calls. Although
UMTS networks offer high capacity, the expected
demand will certainly overcome the available re-
sources. The 3GPP realized the need for broadcasting
and multicasting in UMTS and proposed some en-
hancements on the UMTS Release 6 architecture that
led to the denition of the MBMS framework. MBMS
is a point-to-multipoint service which allows the net-
works resources to be shared [1].
A detailed cost analysis model for the evaluation of
different one to many packet delivery schemes in
UMTS is presented in Reference [2]. The schemes
that the authors consider in the evaluation are the
Broadcast scheme, the Multiple Unicast scheme, and
Multicast scheme. However, in this approach the
authors focus their evaluation in the Core Network
of the UMTS architecture. In Reference [3], a more
detailed analysis of the above-mentioned one to many
delivery schemes is presented. In this work, the
authors consider different transport channels for the
transmission of the data over the UTRAN interfaces.
Both these works do not take into account the evalua-
*Correspondence to: Christos Bouras, Research Academic Computer Technology Institute, N. Kazantzaki str, GR-26500
Patras, Greece.
y
E-mail: bouras@cti.gr
Copyright # 2008 John Wiley & Sons, Ltd.
tion of a number of parameters such as the different
cell environments (micro, macro), the power proles
of the transport channels, and nally the selection of
the most efcient transport channel for the transmis-
sion of the data over the UTRAN interfaces.
In this paper, we present an overview of the MBMS
multicast mode of UMTS. We analytically present the
multicast mode of the MBMS and analyze its perfor-
mance in terms of packet delivery cost under various
network topologies and multicast users distributions
both in macrocell and microcell environments. The
analysis of total packet delivery cost takes into ac-
count the paging cost, the processing cost, and the
transmission cost at nodes and links of the topology.
Furthermore, for the evaluation of the scheme, we
consider different transport channels for the transmis-
sion of the multicast data over the Iub and Uu
interfaces. The transport channels, in the downlink,
currently existing in UMTS which could be used to an
MBMS service are the Dedicated Channel (DCH), the
Forward Access Channel (FACH), and the High-
Speed Downlink Shared Channel (HS-DSCH). How-
ever, in our analysis we will focus on the DCH and
FACH transport channels. The fundamental factor that
determines the transmission cost over the air (Uu
interface) is the amount of Node Bs transmission
power that should be allocated when using each one of
these transport channels. DCH and FACH have dif-
ferent characteristics in terms of power control. Thus,
we present an extended analysis of Node Bs power
consumption in order to dene the exact telecommu-
nication cost introduced by the Iub and Uu interfaces
during the MBMS multicast transmission.
Finally, we propose a switching point scheme for
the efcient radio bearer selection in order to mini-
mize total packet delivery cost. This scheme actually
constitutes a contribution to the MBMS Counting
Mechanism [1]. MBMS counting mechanism exam-
ines whether it is more economical to transmit the
multimedia services in point-to-point (PTP) or point-
to-multipoint (PTM) mode. This mechanism evaluates
whether it is preferable to use dedicated resources
(multiple DCHs) or common resources (a single
FACH). The criteria for the decision of this switching
point should be based on the downlink radio resource
efciency.
This paper is structured as follows. In Section 2, we
provide an overview of the UMTS in packet switched
domain. Section 3 presents the MBMS framework of
UMTS. In Section 4, we present a cost analysis
method for the evaluation of the MBMS multicast
mode. Following this, Section 5 provides important
aspects of power control in MBMS, while Section 6
presents some numerical results. Finally, some con-
cluding remarks and planned next steps are briey
described.
2. Overview of UMTS
and MBMS Architecture
UMTS network is split into two main domains: the
User Equipment (UE) domain and the Public Land
Mobile Network (PLMN) domain. The UE domain
consists of the equipment employed by the user to
access the UMTS services. The PLMN domain con-
sists of two land-based infrastructures: the Core Net-
work (CN) and the UTRAN (Figure 1). The CN is
responsible for switching/routing voice and data con-
nections, while the UTRAN handles all radio-related
Fig. 1. UMTS and MBMS Architecture.
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
functionalities. The CN is logically divided into two
service domains: the Circuit-Switched (CS) service
domain and the Packet-Switched (PS) service domain
[4,5]. The PS portion of the CN in UMTS consists of
two kinds of General Packet Radio Service (GPRS)
Support Nodes (GSNs), namely Gateway GSN
(GGSN) and Serving GSN (SGSN) (Figure 1).
SGSN is the centerpiece of the PS domain. It provides
routing functionality which interacts with databases
(like Home Location Register (HLR)) and manages
many Radio Network Controllers (RNCs). SGSN is
connected to GGSN via the Gn interface and to RNCs
via the Iu interface. GGSN provides the interconnec-
tion of UMTS network (through the Broadcast
MulticastService Center) with other Packet Data
Networks (PDNs) like the Internet [5].
UTRAN consists of two kinds of nodes: the rst is
the RNC and the second is the Node B. Node B con-
stitutes the base station and provides radio coverage to
one or more cells (Figure 1). Node B is connected to
the User Equipment (UE) via the Uu interface (based
on the W-CDMA technology) and to the RNC via the
Iub interface. One RNC with all the Node Bs connected
to it is called Radio Network Subsystem (RNS).
In the UMTS PS domain, the cells are grouped into
Routing Areas (RAs), while the cells in a RA are
further grouped into UTRAN Registration Areas
(URAs). The mobility-management activities for a
UE are characterized by two nite state machines: the
Mobility Management (MM) and the Radio Resource
Control (RRC). The Packet MM (PMM) state ma-
chine for the UMTS PS domain is executed between
the SGSN and the UE for CN-level tracking, while the
RRC state machine is executed between the UTRAN
and the UE for UTRAN-level tracking. After the UE is
attached to the PS service domain, the PMM state
machine is in one of the two states: PMM idle and
PMM connected. In the RRC state machine, there are
three states: RRC idle mode, RRC cell-connected
mode, and RRC URA connected mode [6].
3GPP is currently standardizing the Multimedia
Broadcast/Multicast Service. Actually, the MBMS is
an IP datacast type of service, which can be offered
via existing GSM and UMTS cellular networks. It
consists of a MBMS bearer service and a MBMS user
service. The latter represents applications, which offer
for example multimedia content to the users, while the
MBMS bearer service provides methods for user
authorization, charging, and Quality of Service im-
provement to prevent unauthorized reception. The
major modication in the existing GPRS platform is
the addition of a new entity called Broadcast Multi-
castService Center (BM-SC). Figure 1 presents the
architecture of the MBMS. The BM-SC communi-
cates with the existing UMTS GSM networks and the
external Public Data Networks [7,8].
Three new logical channels are considered for PTM
transmission of MBMS: MBMS point-to-multipoint
Control Channel (MCCH), MBMS point-to-multi-
point Scheduling Channel (MSCH), and MBMS
point-to-multipoint Trafc Channel (MTCH). These
logical channels are mapped on FACH. In case of PTP
transmission Dedicated Trafc Channel (DTCH) and
Dedicated Control Channel (DCCH) are used and are
mapped on the dedicated channel, DCH [1].
3. Description of the MBMS
Multicast Mode
In this section, we present an overview of the multi-
cast mode of the MBMS framework. Figure 2 shows a
subset of a UMTS network. In this architecture, there
are two SGSNs connected to a GGSN, four RNCs, and
twelve Node Bs. Furthermore, eleven members of a
multicast group are located in six cells. The BM-SC
acts as the interface towards external sources of trafc
[8]. In the analysis presented, we assume that a data
stream that comes from an external PDN through BM-
SC must be delivered to the 11 UEs as illustrated in
Figure 2.
The analysis presented in the following paragraphs
covers the forwarding mechanism of the data packets
between the BM-SC and the UEs (Figure 2). Regard-
ing the transmission of the packets over the Iub and
Uu interfaces, it may be performed on common (e.g.,
Forward Access ChannelFACH) or dedicated (Dedi-
cated ChannelDCH) channels. As presented in Re-
ference [9], the transport channel that the 3GPP
decided to use as the main transport channel for
point-to-multipoint MBMS data transmission is the
FACH with turbo coding and QPSK modulation at a
constant transmission power. DCH is a point-to-point
channel and, hence, it suffers from the inefciencies
of requiring multiple DCHs to carry the data to a
group of users. However, DCH can employ fast
closed-loop power control and soft handover mechan-
isms and generally is a highly reliable channel [5,10].
With multicast, the packets are forwarded to those
Node Bs that have multicast users. Therefore, in
Figure 2, the Node Bs 2, 3, 5, 7, 8, 9 receive the
multicast packets issued by the BM-SC. We briey
summarize the ve steps occurred for the delivery of
the multicast packets. Firstly, the BM-SC receives a
RADIO BEARER SELECTION FOR MULTICASTING OVER UMTS
Copyright # 2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
multicast packet and forwards it to the GGSN that has
registered to receive the multicast trafc. Then, the
GGSN receives the multicast packet and by querying
its multicast routing lists, it determines which SGSNs
have multicast users residing in their respective ser-
vice areas. In Figure 2, the GGSN duplicates the
multicast packet and forwards it to the SGSN1 and
the SGSN2 [11]. Then, both destination SGSNs re-
ceive the multicast packets and, having queried their
multicast routing lists, determine which RNCs are to
receive the multicast packets. The destination RNCs
receive the multicast packet and send it to the Node Bs
that have established the appropriate radio bearers for
the multicast application. In Figure 2, these are Node
B2, B3, B5, B7, B8, and B9. The multicast users
receive the multicast packets on the appropriate radio
bearers, either by point-to-point channels transmitted
to individual users separately or by common channels
transmitted to all members in the cell [11].
4. Cost Analysis of the MBMS
Multicast Mode
4.1. General Assumptions
We consider a subset of a UMTS network consisting
of a single GGSN and N
SGSN
SGSN nodes connected
to the GGSN. Furthermore, each SGSN manages a
number of N
ra
RAs. Each RA consists of a number of
N
rnc
RNC nodes, while each RNC node manages a
number of N
ura
URAs. Finally, each URA consists of
N
nodeb
cells. The total number of RAs, RNCs, URAs
and cells are:
N
RA
N
SGSN
N
ra
1
N
RNC
N
SGSN
N
ra
N
rnc
2
N
URA
N
SGSN
N
ra
N
rnc
N
ura
3
N
NODEB
N
SGSN
N
ra
N
rnc
N
ura
N
nodeb
4
The total transmission cost for packet deliveries
including paging is considered as the performance
metric. Furthermore, the cost for paging is differen-
tiated from that cost for packet deliveries. We make a
further distinction between the processing costs at
nodes and the transmission costs on links, both for
paging and packet deliveries. As presented in Refer-
ence [12] and analyzed in Reference [2], we assume
that there is a cost associated with each link and each
node of the network, both for paging and packet
Fig. 2. Packet delivery in UMTS.
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
deliveries. For the analysis, we apply the following
notations:
D
gs
Tx cost of packet delivery between GGSN
and SGSN
D
sr
Tx cost of packet delivery between SGSN
and RNC
D
rb
Tx cost of packet delivery between RNC and
Node B
D
DCH
Tx cost of packet delivery over Uu with DCHs
D
FACH
Tx cost of packet delivery over Uu with
FACHs
S
sr
Tx cost of paging between SGSN and RNC
S
rb
Tx cost of paging between RNC and Node B
S
a
Tx cost of paging over the air
p
gM
Processing cost of multicast packet delivery
at GGSN
p
sM
Processing cost of multicast packet delivery
at SGSN
p
rM
Processing cost of multicast packet delivery
at RNC
p
b
Processing cost of packet delivery at Node B
a
s
Processing cost of paging at SGSN
a
r
Processing cost of paging at RNC
a
b
Processing cost of paging at Node B
The total number of the multicast UEs in the net-
work is denoted by N
UE
. For the cost analysis, we
dene the total packets per multicast session as N
p
.
Since network operators will typically deploy an IP
backbone network between the GGSN, SGSN, and
RNC, the links between these nodes will consist of
more than one hop. Additionally, the distance between
the RNC and Node B consists of a single hop (l
rb
1).
In the presented analysis, we assume that the distance
between GGSN and SGSN is l
gs
hops, while the
distance between the SGSN and RNC is l
sr
hops.
We assume that the probability that a UE is in the
PMM detached state is P
DET
, the probability that a UE
is in the PMM idle/RRC idle state is P
RA
, the prob-
ability that a UE is in the PMM connected/RRC URA
connected state is P
URA
, and nally the probability
that a UE is in the PMM connected/RRC cell-con-
nected state is P
cell
.
In the remainder of this section, we describe a
method that models the multicast user distribution in
the network. In particular, we present a probabilistic
method that calculates the number of multicast users
in the network (N
UE
), the number of SGSNs that serve
multicast users (n
SGSN
), the number of RNCs that
serve multicast users (n
RNC
), and nally the number of
Node Bs that serve multicast members (n
NODEB
).
As introduced in Reference [13] and analyzed in
Reference [2], we classify the RAs into L
RA
categories
in order to create an asymmetric topology. For
1 i L
RA
there are N
RA
i
RAs of class i. Therefore,
the total number of RAs within the network is
N
RA

L
RA
i1
N
RA
i
.
Suppose that the distribution of the multicast users
among the classes of RAs follows the Poisson dis-
tribution with
RA
i
where 1 i L
RA
. In gen-
eral, the probability that exactly k multicast users
reside in the RAs of class i is calculated from the
following equation:
p k;
RA
i
_ _

RA
i

RA
i
_ _
k
k!
5
Thus, the probability none of the RAs of class i
serves multicast users is p0;
RA
i
e

RA
i
, which
in turn means that the probability at least one multi-
cast user is served by the RAs of class i is
p 1 p0;
RA
i
1 e

RA
i
.
Since every class i consists of N
RA
i
RAs, the total
number of the RAs in the class i, that serve multicast
users is N
RA
i
1 e

RA
i
. Thus, the total number of
the RAs of every class that serve multicast users is
n
RA

L
RA
i1
N
RA
i
1 e

RA
i
_ _
6
where
RA
i
represents the number of multicast users
for the N
RA
i
RAs of class i.
If there are n
RA
RAs that are serving multicast
users, the probability that an SGSN does not have any
such RA is
p
SGSN

N
RA
N
ra
n
RA
_ __
N
RA
n
RA
_ _
;
0;
_
_
_
n
RA
N
RA
N
ra
otherwise
7
Based on Equation (7), the total number of SGSNs
that are serving multicast users can be calculated as
follows: n
SGSN
N
SGSN
1 p
SGSN
.
The total number of multicast users in the network is
N
UE

L
RA
i1
N
RA
i

i
8
where
i
is the number of multicast users in a RA of
class i.
RADIO BEARER SELECTION FOR MULTICASTING OVER UMTS
Copyright # 2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
As in Reference [2], we assume that all RNCs
within a service area of class i have the same multicast
population distribution density as in the RA case.
Based on a uniform density distribution within a
single RA, the multicast population of an RNC within
the service area of a class i RA is
RNC
i

RA
i
=N
rnc
.
The total number of RNCs of class i is N
RNC
i

N
RA
i
N
rnc
.
Assuming that the number of RA categories is equal
to the number of RNC categories (L
RNC
L
RA
), the
total number of RNCs that serve multicast users is
n
RNC

L
RNC
i1
N
RNC
i
1 e

RNC
i
_ _
9
The same are applied to the cells within the service
area of an RNC. The average number of multicast
users for a single cell of class i is
B
i

RNC
i
=
N
ura
N
nodeb
.
The number of Node Bs belonging to class i is
N
B
i
N
RNC
i
N
ura
N
nodeb
. Assuming that the number
of the RNC categories is equal to the number of the
Node B categories (L
RNC
L
NODEB
), the total number
of Node Bs that serve multicast users is
n
NODEB

L
NODEB
i1
N
B
i
1 e

B
i
_ _
10
4.2. Cost Analysis of the Multicast Mode
In the multicast scheme, the multicast group manage-
ment is performed at the BM-SC, GGSN, SGSN, and
RNC and multicast tunnels are established over the
Gn and Iu interfaces. It is obvious that the cost of a
single packet delivery to a multicast user depends on
its MM and RRC state.
If the multicast member is in the PMM connected/
RRC cell-connected state, then there is no need for
any paging procedure either from the SGSN or from
the serving RNC. In this case, the packet delivery cost
is derived from Equation (11). It has to be mentioned
that this quantity does not include the cost for the trans-
mission of the packets over the Iub and Uu interfaces,
since this cost depends rstly on the number of
multicast users and secondly on the transport channel
used for data transmission.
C
cell
p
gM
D
gs
p
sM
D
sr
p
rM
11
If the multicast member is in the PMM connected/
RRC URA connected state, then the RNC must rst
page all the cells within the URA in which mobile
users reside and then proceed to the data transfer.
After the subscriber receives the paging message from
the RNC, it returns to the RNC its cell ID. The cost for
paging such a multicast member is
C
URA
N
nodeb
S
rb
a
b
S
a
S
a
a
b
S
rb
a
r
12
If the multicast member is in the PMM idle/RRC
idle state, the SGSN only stores the identity of the RA
in which the user is located. Therefore, all cells in the
RA must be paged. The cost for paging such a multi-
cast member is
C
RA
N
rnc
S
sr
a
r
N
rnc
N
ura
N
nodeb

S
rb
a
b
S
a
S
a
a
b
S
rb
a
r
S
sr
a
s
13
After the paging procedure, the RNC stores the
location of any UE at a cell level. In multicast, the
SGSN and the RNC forward a single copy of each
multicast packet to those RNCs or Node Bs respec-
tively that are serving multicast users. After the
correct multicast packet reception at the Node Bs
that serve multicast users, the Node Bs transmit the
multicast packets to the multicast users via common
or dedicated transport channels. The total cost for the
multicast scheme is derived from the following equa-
tion where n
SGSN
, n
RNC
, n
NODEB
represent the number
of SGSNs, RNCs, Node Bs respectively that serve
multicast users.
Ms p
gM
n
SGSN
D
gs
p
sM
_ _
n
RNC
D
sr
p
rM
Y
_
N
p
P
RA
C
RA
P
URA
C
URA
N
UE
D
packet delivery
D
paging
14
where
Y
n
NODEB
D
rb
p
b
D
FACH
; if channel FACH
N
UE
D
rb
p
b
D
DCH
; if channel DCH
_
D
packet delivery
p
gM
n
SGSN
D
gs
p
sM
_ _
n
RNC
D
sr
p
rM
Y
_
N
p
D
paging
P
RA
C
RA
P
URA
C
URA
N
UE
Parameter Y represents the multicast cost for the
transmission of the multicast data over the Iub and Uu
interfaces. This cost depends mainly on the distribu-
tion of the multicast group within the UMTS network
and secondly on the transport channel that is used.
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
D
DCH
and D
FACH
represent the cost over the Uu
interface. More specically, D
FACH
represents the
cost of using a FACH channel to serve all the multicast
users residing in a specic cell while D
DCH
represents
the cost of using a single DCH to transmit the multi-
cast data to a single multicast user of the network.
Regarding the cost over the Iub interface, in case we
use the FACH as transport channel, each multicast
packet send once over the Iub interface and then the
packet is transmitted to the UEs that are served by the
corresponding Node B. On the other hand, in case we
use DCHs for the transmission of the multicast pack-
ets over the Iub each packet is replicated over the Iub
as many times as the number of multicast users that
the corresponding Node B serves.
5. Power Control in MBMS
In this section, some important issues regarding the
power control of the downlink transport channels
(DCH and FACH) are analyzed. This analysis is
performed in order to determine, as will be presented
in the next section, the exact values of parameters D
DCH
and D
FACH
, appearing in Equation (14). It is recalled
that the main factor that determines the MBMS trans-
mission cost over the Uu interface is the amount of
Node Bs transmission power that should be allocated
when using one of these transport channels.
Power control is one of the most important aspects
in MBMS due to the fact that Node Bs transmission
power is a limited resource and must be shared among
all MBMS users in a cell. Power control is essential in
order to minimize the transmitted power, thus avoid-
ing unnecessary high power levels and eliminating
intercell interference. The main requirement is to
make an efcient overall usage of the radio resources:
this makes the common channel, FACH, the favorite
choice, since many users can access the same resource
at the same time. However, other crucial factors such
as the number of users belonging to the multicast
group and their distance from the serving Node B, the
type of service provided, and the QoS requirements
(represented by E
b
=N
0
targets) affect the choice of the
most efcient transport channel in terms of power
consumption.
On the point-to-point downlink transmissions,
where multiple DCHs are used, fast power control is
used to maintain the quality of the each link and thus
to provide a reliable connection for the receiver to
obtain the data with acceptable error rates. Transmit-
ting with just enough power to maintain the required
quality for the link also ensures that there is minimum
interference affecting the neighboring cells. Transmis-
sion power allocated for all MBMS users in a cell that
are served by multiple DCHs is variable. It mainly
depends on the number of UEs, their location in the
cell (close to the Node B or at cell edge), the required
bit rate of the MBMS session, and the experienced
signal quality E
b
=N
0
for each user. Equation (15)
calculates the Node Bs total transmission power
required for the transmission of the data to n users
when multiple DCHs are used [14].
P
T

P
P

n
i1
P
N
x
i

W
E
b
=N
0

i
R
b;i
p
L
p;i
1

n
i1
p
W
E
b
=N
0

i
R
b;i
p
15
where P
T
the total transmission power for all the DCH
users in the cell, P
P
the power devoted to common
control channels, L
p;i
refers to the path loss for user i,
R
b;i
the bit rate for user i, W the bandwidth, P
N
the
background noise, p the orthogonality factor (0:per-
fect orthogonality), and E
b
=N
0

i
is the signal energy
per bit divided by noise spectral density. Parameter x
i
is the intercell interference observed by user i given as
a function of the transmitted power by the neighboring
cells P
Tj
, j 1, . . . ,K and the path loss from this user
to the jth cell L
ij
. More specically [14]:
x
i

K
j1
P
Tj
L
ij
16
In contrast, in point-to-multipoint downlink trans-
missions, a single FACH is established and essentially
transmits at a xed power level since fast power control
is not supported in this channel. A FACH channel must
be received by all UEs throughout the cell. Conse-
quently, the xed power should be high enough to
ensure the requested QoS in the whole coverage area
of the cell, irrespective of the UEs location. FACH
power efciency depends on maximizing diversity as
power resources are limited. Diversity can be obtained
by using a longer TTI, e.g., 80 ms instead of 20 ms, to
provide time diversity against fast fading (fortunately,
MBMS services are not delay sensitive) and the use of
combining transmissions from multiple cells to obtain
macro diversity [15].
Power aspects of MBMS are investigated sepa-
rately for macro and microcell environments. The
amount of intercell interference is lower in microcells
where street corners isolate the cells more strictly than
RADIO BEARER SELECTION FOR MULTICASTING OVER UMTS
Copyright # 2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
in macrocells. Moreover, in microcells there is less
multipath propagation, and thus a better orthogonality
of the downlink codes. On the other hand, less multi-
path propagation gives less multipath diversity, and
therefore a higher E
b
=N
0
requirement in the downlink
in micro than in macrocells is assumed [5]. The basic
simulation parameters are presented in Tables I and II
[1619].
6. Results
6.1. Simulation and Evaluation Parameters
In this section, we present the evaluation parameters
regarding the MBMS multicast mode. We consider
different cell congurations, different user distribu-
tions, and nally, different transport channels for the
transmission of the multicast data over the UTRAN
interfaces. Therefore, we assume a general network
topology, with N
SGSN
10, N
ra
10, N
rnc
10,
N
ura
5, and N
nodeb
5.
The packet transmission cost (D
xx
) in any segment
of the UMTS network depends on two parameters: the
number of hops between the edge nodes of this net-
work segment and the capacity of the link of the
network segment. This means that D
gs
l
gs
/k
gs
, D
sr

l
sr
/k
sr
, and D
rb
l
rb
/k
rb
. Parameter k
xx
represents the
prole of the corresponding link between two UMTS
network nodes. More specically, in the high capacity
links at the CN, the values of k
xx
are greater than the
corresponding values in the low capacity links at
UTRAN. For the cost analysis and without loss of
generality, we assume that the distance between the
GGSN and SGSN is 8 hops, the distance between
SGSN and RNC is 4 hops, and the distance between
RNC and Node B is 1 hop. The above parameters as
well as the values of the k
xx
are presented in detail in
Table III. Regarding the transmission cost of paging
(S
xx
) in the segments of the UMTS network, it is
calculated in a similar way as the packet transmission
cost (D
xx
). More specically, S
xx
is a fraction of the
calculated transmission cost (D
xx
) and in our case we
assume that it is three times smaller than D
xx
.
As we can observe from the equations of the
previous section, the costs of the schemes depend on
a number of other parameters. Thus, we have to
estimate the value of these parameters. The chosen
values of the parameters are presented in Table IV.
At this point, we have to mention that since the
nodes that are responsible for the forwarding of the
multicast packets are the GGSN, SGSN, and the RNC,
we consider a lower packet processing cost in the
Node B than the corresponding costs in the GGSN,
SGSN, and RNC since some overhead is needed in the
above-mentioned three nodes in order to maintain
the routing lists required for the packet forwarding
in the multicast scheme (Table IV).
Table I. Macrocell simulation assumptions.
Parameter Value
Cellular layout Hexagonal grid
Number of neighboring cells 18
Sectorization 3 sectors/cell
Site to site distance 1 km
Cell radius 0.577 km
Maximum BS Tx power 20 W (43 dBm)
Other BS Tx power 5 W (37 dBm)
Common channel power 1 W (30 dBm)
Propagation model Okumura Hata
Multipath channel Vehicular A (3 km/h)
Orthogonality factor 0.5
(0: perfect orthogonality)
E
b
=N
0
target 5 dB
FACH Tx power 4 W (32 kbps service)
(no STTD, 95% coverage) 7.6 W (64 kbps service)
15.8 W (128 kbps service)
Table II. Microcell simulation assumptions.
Parameter Value
Cellular layout Manhattan grid
Number of cells 72
Block width: Road width: 75 m: 15 m: 90 m
Building to building distance
Straight line distance between 360 m (four blocks)
transmitters
Maximum BS Tx power 2 W (33 dBm)
Other BS Tx power 0.5 W (27 dBm)
Common channel power 0.1 W (20 dBm)
Propagation model Walsh-Ikegami
Multipath channel Pedestrian A 3 km/h
Orthogonality factor 0.1
(0: perfect orthogonality)
E
b
=N
0
target 6 dB
FACH Tx power 0.36 W (64 kbps service)
(no STTD, 95% coverage)
Table III. Chosen values for the calculation of transmission costs in
the links.
Link Link Capacity Number of Transmission
factor (k) hops (l) cost (D)
GGSN-SGSN k
gs
0.8 l
gs
8 D
gs
10
SGSN-RNC k
sr
0.7 l
sr
4 D
sr
4/0.7
RNCNode B k
rb
0.5 l
rb
1 D
rb
2
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
Furthermore, we have appropriately chosen the
probabilities P
RA
, P
URA
, and P
cell
More specically,
the probability that a UE is in the PMM idle/RRC idle
state is P
RA
0.6. The probability that a UE is in the
PMM connected/RRC URA connected state is
P
URA
0.2, and the probability that a UE is in the
PMM connected/RRC cell-connected state is
P
cell
0.1. Additionally, there is a probability that
the UE is not reachable by the network and we
consider it to be 0.1.
Regarding the transmission over the Iub and Uu,
DCH and FACH channels are examined. Some im-
portant aspects concerning the power consumption for
these two transport channels were presented analyti-
cally in Section 5. It is recalled that the amount of
Node Bs transmission power that must be allocated
for these two channels is the main parameter that
denes the transmission cost over the air (Uu inter-
face). Parameter D
FACH
represents the cost, over the
Uu interface, of using a FACH channel to serve all the
multicast users. Similarly, parameter D
DCH
represents
the cost, over the Uu interface, of using a single DCH
channel to serve one multicast user.
In our analysis, we calculate in each cell of the
network topology the Node Bs power in the case of
using DCHs or FACH. Then, by comparing these
power values with the total available Node Bs trans-
mission power, we select the appropriate values for
D
DCH
and D
FACH
Obviously the values D
DCH
and
D
FACH
are proportional to the percentage of the
Node Bs transmission power allocated to DCH or
FACH in any cell. The D
DCH
and D
FACH
values are
then used in Equation (14) that calculates the total
telecommunication cost of the MBMS multicast
mode. Furthermore, we assume that the minimum
value that the total D
DCH
per cell and the D
FACH
could
take is the value of 10 since this value is the cost of the
data transmission in the wired link between the GGSN
and the SGSN (D
gs
), and generally the transmission
cost in a wired link is assumed to be lower than that in
a wireless link.
It is true that the performance of the multicast
scheme depends mainly on the conguration of the
UMTS network that is under investigation. In our
analysis, we assume that we have two classes of RAs.
A class i 1 RA has multicast user population of

1
1/ and a class i 2 RA has a multicast user
population of
2
. If ) 1, the class i 1 RA has a
small multicast user population and the class i 2 RA
has a large multicast user population. Let be the
proportion of the class i 1 RAs and (1 ) be
the proportion of the class i 2 RAs [13]. Thus,
the number of class i 1 RAs is N
RA
1
N
RA
and
the number of class i 2 RAs is N
RA
2
1 N
RA
.
Each RA of class i 2 f1; 2g is in turn sub-divided into
N
rnc
RNCs of the same class i and similarly, each RNC
of class i 2 f1; 2g is sub-divided into N
ura
N
nodeb
Node Bs of the same class i. Taking into consideration
the above-mentioned parameters, Equation (8) can be
transformed to Equation (17). It is obvious from
Equation (17) that as decreases and increases
the number of multicast users increases rapidly:
N
UE

2
i1
N
RA
i

i
N
RA
1

1
N
RA
2

2
N
RA

_ _
17
6.2. Telecommunications Cost of the Data
Transmission Over the Uu Interface
In this section, analytical simulation results, distinctly
for the cases of macro and microcell environments,
are presented. Node Bs transmission power levels
when using DCH or FACH channels, for different
simulation parameters, are depicted in each of the
following gures. The aim for this parallel plotting is
to determine the most efcient transport channel, in
terms of power consumption, for the transmission of
the MBMS data.
The effect of UEs location (distance from Node B)
is presented in Figure 3. UEs are assumed to be in
groups, located at the same distance from Node B
each time. When multiple DCHs are used, it is
obvious that the further a UE is from the Node B
the more power is required for the successful delivery
of the MBMS service, both for macrocell and micro-
cell environments. On the other hand, when a FACH
Table IV. Chosen parameters values.
S
sr
S
rb
S
a
p
gM
p
sM
p
rM
p
b
a
s
a
r
a
b
P
RA
P
URA
P
cell
4/2.1 2/3 4/3 2 2 2 1 1 1 1 0.6 0.2 0.1
RADIO BEARER SELECTION FOR MULTICASTING OVER UMTS
Copyright # 2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
channel is used, transmission power is kept constant
(irrespective of UEs location) at a power level that is
high enough to serve the UE with the greater distance
from Node B. In Figure 3, FACH Tx power is set to a
value that provides 95% coverage (cell edge). For
smaller coverage areas FACH Tx power could be set
in lower levels.
Figure 4 reects the impact of QoS requirements
(E
b
=N
0
) on Node B transmission power. As expected,
the higher the E
b
=N
0
parameter is the more power is
required when transmitting multicast data with multi-
ple DCHs.
Figure 5 depicts the impact of the MBMS bit rate on
Node B transmission power. When multiple DCHs are
used, increased MBMS bit rates result in higher Node
B transmission power. Similarly, in the case of a
FACH channel, more power needs to be transmitted
when providing higher MBMS bit rates. FACH trans-
mission power levels for various bit rates are depicted
for a macrocell environment (Figure 5a), while for a
microcell environment (Figure 5b) we consider only a
64 kbps service.
Another crucial factor that has to be taken into
account when selecting the most efcient transport
channel is the transmission power of neighboring
cells, expressed by the parameter P
Tj
in Equation
(16). Figure 6 depicts the impact of this factor under
the simplifying assumption that all neighboring Node
Bs transmit at the same power levels. Of course, it is
rather impossible that all neighboring Node Bs trans-
mit at the same power level, but this is assumed here
for better understanding of this parameters signicant
Fig. 3. Tx power versus. distance (a) Macrocell, (b) Microcell.
Fig. 4. Tx power versus. Eb/No (a) Macrocell, (b) Microcell.
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
impact. Higher transmission power of neighboring
Node Bs increases intercell interference, leading in
turn the examined Node B to increase its transmission
power in order to meet the MBMS service demands.
From the above gures that present the cost of
MBMS transmission over the Uu interface, useful
information about the switching point between
point-to-point transmission (multiple DCHs) and
point-to-multipoint transmission (a single FACH)
can be extracted. Actually, a power-based switching
point scheme can be employed in order to minimize
Node Bs transmission power, thus minimizing the
cost for the transmission of the multicast data over the
air. The transport channel that requires less power
resources is selected. For instance, from Figure 3a in
the case of a macrocell, it can be seen that for a
64 kbps MBMS service, E
b
=N
0
target 5 dB, 95%
coverage and neighboring Node Bs transmitting at
5 W an efcient switching point should be 9 UEs.
Furthermore, in the case of a microcell, for a 64 kbps
MBMS service, E
b
=N
0
target 6 dB, 95% coverage and
neighboring Node Bs transmitting at 0.5 W the
switching point should be 7 UEs (Figure 3b). This
means that, e.g., for a macrocell environment, for 9
UEs and above a FACH should be used, while for less
than 9 UEs the use of multiple DCHs is the most
efcient choice.
6.3. Total Telecommunication Cost
In Figure 7 the total costs for the multicast mode using
different transport channels and cell environments in
function of are presented. From these plots, we can
see that the costs decrease as increases. This occurs
Fig. 5. Tx power versus. bit rate (a) Macrocell, (b) Microcell.
Fig. 6. Tx power versus. neighboring cells Tx power (a) Macrocell, (b) Microcell.
RADIO BEARER SELECTION FOR MULTICASTING OVER UMTS
Copyright # 2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
because as increases the number of RAs with no
multicast users increases and hence the multicast
users are located in a small number of RAs.
More specically, in Figure 7a, the cost in case we
use multiple DCHs is smaller than the cost in case we
use a FACH channel both in macro and microenviron-
ments. This occurs because the small value of results
in a reduced number of UEs in the network and hence
the DCH is more efcient for the data transmission in
terms of total cost. The opposite occurs in Figure 7b
where the value of is increased, which means that the
number of UEs is also increased. Therefore, the use of
DCHs is inefcient for the transmission of the data
over the Iub and Uu interfaces while the FACH is the
most suitable transport channel in terms of total cost.
In Figure 8, the total costs using different transport
channels and cell environments in the function of are
presented. We choose a small value for the parameter
because the multicast mode becomes efcient when
there is an increased density of UEs in the network
[3]. Therefore, a value of 0.1 is chosen which
means that there are many RAs in the network with a
great number of multicast users in these. From Figure
8, it is clear that as parameter increases (which
means that the number of multicast users increases),
the total cost for all cases increases too. However, the
increase in total cost for DCHs is greater than that of
FACH due to the fact that a DCH is a point-to-point
channel and strongly depends on the number of multi-
cast users.
More specically, in Figure 8, we observe that for
small values of , the total cost using DCHs is small
because there is a small number of UEs in the net-
work, while for bigger values of , which implies
bigger number of UEs, the total cost using DCHs
overcomes the cost of using FACH. Thus, for small
values of the use of DCHs is more efcient while for
bigger values of , the use of FACH is more appro-
priate. The simulation parameters for DCH and FACH
transport channels used for the plotting of Figures 7
and 8 are taken from the previous section. More
specically, a 64 kbps MBMS service and 95% cell
coverage are assumed.
An important notice regarding the switching point
between point-to-point and point-to-multipoint trans-
port channels should be mentioned at this point. From
Figure 8 the switching point between multiple DCHs
and a single FACH, in terms of total transmission cost,
is 6 UEs (or 1500) for a macrocell and 3 UEs (or
750) for a microcell. However, from the previous
section, when only the cost over the Uu interface
Fig. 7. Total cost in function of with (a) 300, (b) 3000.
Fig. 8. Total cost in function of , 0.1.
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
(Node Bs total transmission power) was taken into
account, it was shown that for the same simulation
parameters, the switching point should be 9 UEs for a
macrocell and 7 UEs for a microcell. Consequently, it
is obvious that a reduction in the switching point
levels is taking place when considering the total
transmission cost of an MBMS session. This reduction
is caused by the additional cost introduced by the Iub
interface, representing the transmission cost of packet
delivery between RNC and Node B. Recall from
Equation (14) that computes the total cost of the
multicast scheme, that the parameter Y represents
the multicast cost for the transmission of the multicast
data over the Iub and Uu interfaces. When a FACH
transport channel is used each multicast packet is sent
once over the Iub, while when multiple DCHs are used
each packet is replicated over the Iub as many times as
the number of multicast users. The cost added from
Iub is not negligible and depends on the link capacity
which is, however, operator dependent. For the simu-
lations presented above, the link capacity factor was
set to k
rb
0.5. For greater values of k
rb
, the switching
points converge to the values presented in Figures 36.
From the above observation, it is clear that the
selection of the appropriate radio bearer for the multi-
cast data transmission strongly depends on the cost
added by the Iub interface. The Node Bs transmission
power should not be the only criterion for the selection
of an efcient transport channel, but the total trans-
mission cost (including the Iub cost) should always be
taken into account.
7. Conclusions and Future Work
In this paper we presented an overview of the MBMS
multicast mode of UMTS. We investigated the per-
formance of the multicast mode of the MBMS in
terms of packet delivery cost through an analytic
theoretical model and by simulations based on this
model. The investigations were made assuming var-
ious network topologies, cell environments, and mul-
ticast users distributions. In addition, we examined
the DCH and FACH transport channels in terms of
data transmission cost over the Iub and Uu interfaces.
Finally, we presented a scheme for the efcient
selection of a switching point between point-to-point
(multiple DCHs) and point-to-multipoint (a single
FACH) transmissions that minimizes the total trans-
mission cost of multicast data.
The step that follows this work is to examine the
impact of the HS-DSCH on the total transmission cost
of the multicast mode of MBMS. HS-DSCH is a
shared channel, introduced in the Release 5 of
UMTS, and can be used as a transport channel for
the transmission of the MBMS data over the Iub and
Uu interfaces. HSDPA is a key technology for MBMS
as it improves the MBMS performance and increases
bit rate speeds to support new MBMS services [20].
References
1. 3GPP TS 25.346 V7.1.0. Technical Specication Group Radio
Access Network; Introduction of the Multimedia Broadcast
Multicast Service (MBMS) in the Radio Access Network
(RAN), Stage 2 (Release 7). 2006.
2. Rummler R, Chung Y, Aghvami H. Modeling and Analysis of
an Efcient Multicast Mechanism for UMTS. IEEE Transac-
tions on Vehicular Technology 2005; 54(1): 350365.
3. Alexiou A, Bouras C. Multicast in UMTS: Evaluation and
Recommendations. Wireless Communications and Mobile
Computing 2008; 8(4): 463481.
4. 3GPP TS 23.060 V7.0.0. Technical Specication Group Ser-
vices and System Aspects; General Packet Radio Service
(GPRS); Service description; Stage 2 (Release 7). 2006.
5. Holma H, Toskala A. WCDMA for UMTS: Radio Access for
Third Generation Mobile Communications. John Wiley &
Sons: Chichester, UK, 2004.
6. Yang S, Lin Y. Performance evaluation of location manage-
ment in UMTS. IEEE Transactions on Vehicular Technology
2003; 52(6): 16031615.
7. 3GPP TS 22.146 V7.1.0. Technical Specication Group Ser-
vices and System Aspects; Multimedia Broadcast/Multicast
Service; Stage 1 (Release 7). 2006.
8. 3GPP TS 23.246 V6.9.0. Technical Specication Group Ser-
vices and System Aspects; Multimedia Broadcast/Multicast
Service (MBMS); Architecture and functional description
(Release 6). 2005.
9. 3GPP, TR 23.846 v6.1.0. Technical Specication Group Ser-
vices and System Aspects; Multimedia Broadcast/Multicast
Service; Architecture and functional description (Release 6).
10. Boni A, Launay E, Mienville T, Stuckmann P. Multimedia
Broadcast Multicast ServiceTechnology Overview and Ser-
vice Aspects. 5th IEE International Conference on 3G Mobile
Communication Technologies (3G 2004), London, UK, 634
638.
11. Alexiou A, Antonellis D, Bouras C, Papazois A. An Efcient
Multicast Packet Delivery Scheme for UMTS. The 9th ACM/
IEEE International Symposium on Modeling, Analysis and
Simulation of Wireless and Mobile Systems (MSWiM 2006),
Torremolinos, Malaga, Spain, 2006; 147150.
12. Ho J, Akyildiz I. Local anchor scheme for reducing signaling
costs in personal communications networks. IEEE/ACM Trans-
actions on Networking 1996; 4(5): 709725.
13. Lin Y. A multicast mechanism for mobile networks. IEEE
Communication Letters 2001; 5(11): 450452.
14. Romero J, Sallent O, Agusti R, Diaz-Guerra M. Radio Re-
source Management Strategies in UMTS. John Wiley & Sons:
Chichester, UK, 2004.
15. Parkvall S, Englund E, Lundevall M, Torsner J. Evolving 3G
Mobile Systems: Broadband and Broadcast Services in
WCDMA. IEEE Communications Magazine 2006; 44(2): 3036.
16. 3GPP TR 101.102 V3.2.0. Universal Mobile Telecommunica-
tions System (UMTS); Selection procedures for the choice of
radio transmission technologies of the UMTS (UMTS 30.03
version 3.2.0).
RADIO BEARER SELECTION FOR MULTICASTING OVER UMTS
Copyright # 2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm
17. Czerepinski P, Chapman T, Krause J. Coverage and planning
aspects of MBMS in UTRAN. 5th IEE International Confer-
ence on 3G Mobile Communication Technologies (3G 2004),
London, UK, 2004; 529533.
18. 3GPP TS 25.803 v6.0.0, Technical Specication Group Radio
Access Network; S-CCPCH performance for MBMS, (Release
6).
19. IST-2003-507607 (B-BONE), Deliverable of the project
(D2.5). Final Results with combined enhancements of the Air
Interface.
20. 3G/UMTS Evolution: towards a new generation of broadband
mobile services, UMTS Forum, White paper, December 2006,
available at: http://www.umts-forum.org.
Authors Biographies
Antonios Alexiou obtained his
Diploma from the Department of
Electrical and Computer Engineering
of the Aristotle University of Thessa-
loniki (Greece). Furthermore, he
obtained his Master Degree and his
Ph.D. from the Computer Engineering
and Informatics Department of Patras
University (Greece). His research
interests include data networks,
mobile telecommunications networks, multicast routing
and group management, radio resource management, radio
network planning and optimization, and wireless networks.
He has published over 40 papers in Journals, International
Conferences, and Book Chapters.
Christos Bouras obtained his
Diploma and Ph.D. from the Depart-
ment Of Computer Engineering and
Informatics of Patras University
(Greece). He is currently a Professor
in the above department. Also he is a
scientic advisor of Research Unit 6 in
Research Academic Computer Tech-
nology Institute (CTI), Patras, Greece.
His research interests include Analysis of Performance of
Networking and Computer Systems, Computer Networks
and Protocols, Telematics and New Services, QoS and
Pricing for Networks and Services, e-Learning Networked
Virtual Environments and WWW Issues. He has extended
professional experience in Design and Analysis of Net-
works, Protocols, Telematics, and New Services. He
has published 300 papers in various well-known refereed
conferences and journals. He is a co-author of eight books in
Greek. He has been a PC member and referee in various
international journals and conferences. He has participated
in R&D projects such as RACE, ESPRIT, TELEMATICS,
EDUCATIONAL MULTIMEDIA, ISPO, EMPLOYMENT,
ADAPT, STRIDE, EUROFORM, IST, GROWTH, and
others. Also he is member of experts in the Greek Research
and Technology Network (GRNET), Advisory Committee
Member to the World Wide Web Consortium (W3C), IEEE
CS Technical Committee on Learning Technologies, IEEE
ComSoc Radio Communications Committee, IASTED
Technical Committee on Education WG6.4 Internet Appli-
cations Engineering of IFIP, ACM, IEEE, EDEN, AACE,
New York Academy of Sciences and Technical Chamber of
Greece.
Vasileios Kokkinos obtained his
diploma from the Physics Department
of the University of Patras, Greece in
October 2003. Next, he was accepted
in the postgraduate program Electro-
nics and Information Processing in
the same department and in March
2006 he obtained his Master Degree.
Currently, he is a Ph.D. student in the
Computer Engineering and Infor-
matics Department. He works in the Research Unit 6 of
Research Academic Computer Technology Institute since
September 2006. His research interests include wireless
networks, third generation mobile telecommunications net-
works, multicast routing and group management, and radio
resource management. He has published eight papers in
well-known refereed conferences.
Evangelos Rekkas obtained his
diploma from the Computer Engineer-
ing and Informatics Department of the
University of Patras (Greece) in Octo-
ber 2007. He is currently an M.Sc.
Student in the Computer Engineering
and Informatics Department. Further-
more, he is working as R&D Compu-
ter Engineer at the Research Unit 6 of
the Research Academic Computer
Technology Institute in Patras
(Greece) since September 2006. His
research interests include Mobile Telecommunications Net-
works, Multicast Routing, Radio Network Planning, and
Resource Management in Cellular Networks. He has pub-
lished nine papers in well-known refereed conferences.
A. ALEXIOU ET AL.
Copyright #2008 John Wiley & Sons, Ltd. Wirel. Commun. Mob. Comput. (2008)
DOI: 10.1002/wcm