Int. J. Wireless and Mobile Computing, Vol. 2, No. 4, 2007
Interworking of 3G cellular networks and
Wei Song and Weihua Zhuang*
Centre for Wireless Communications,
Department of Electrical and Computer Engineering,
University of Waterloo,
Waterloo, ON, Canada N2L 3G1
Wireless Technology Department,
5099 Creekbank Road,
Canada L4W 5N2
Abstract: The Third Generation (3G) cellular networks provide ubiquitous connectivity but low
data rates, whereasWireless LocalArea Networks (WLANs) can offer much higher data rates but
only cover smaller geographic areas. Their complementary characteristics make the integration
of the two networks a promising trend for next-generation wireless networks. With combined
in hotspots. There are many aspects involved in their interworking, such as mobility, security
and Quality of Service (QoS) provisioning. In this paper, we present a survey of most recent
interworking mechanisms proposed in the literature, and outline some important open issues to
achieve seamless integration.
Keywords: 3G/WLAN interworking; tight/loose coupling; mobility management; quality of
services; QoS provisioning.
Reference to this paper should be made as follows: Song, W., Zhuang, W. and Saleh, A. (2007)
Vol. 2, No. 4, pp.237–247.
Biographical notes: Wei Song received a BSc in Electrical Engineering from Hebei
University, China, in 1998, an MSc in Computer Science from Beijing University of Posts and
Telecommunications, China, in 2001 and a PhD in Electrical Engineering from the University
of Waterloo, Canada, in 2007. Currently, she is a PostDoctoral Fellow in the Department of
Electrical and Computer Engineering, at the University ofWaterloo, Canada. Her current research
interests include resource allocation and Quality-of-Service (QoS) provisioning for the integrated
cellular networks and Wireless Local Area Networks (WLANs). She received the Best Student
PaperAward from IEEE WCNC’07.
Weihua Zhuang received a PhD in Electrical Engineering from the University of New Brunswick,
Canada. Since October 1993, she has been with the Department of Electrical and Computer
Engineering, University of Waterloo, Canada, where she is co-author of the textbook Wireless
Communications and Networking (Prentice Hall, 2003). Her current research interests include
multimedia wireless communications, wireless networks and radio positioning. She received the
Premier’s Research Excellence Award in 2001 from the Ontario Government for demonstrated
excellence of scientific and academic contributions, the Outstanding PerformanceAward in 2005
and 2006 from the University of Waterloo, and the Best Paper Awards from IEEE WCNC’07,
IEEE ICC’07 and Qshine’07. She is the Editor-in-Chief of IEEE Transactions on Vehicular
Technology and an Editor of IEEE Transactions onWireless Communications, EURASIP Journal
on Wireless Communications and Networking and International Journal of Sensor Networks.
Aladdin Saleh received a PhD in Electrical Engineering from the University of London, England,
in 1984. Since then, he has held several positions both in the academia and the industry. In 1998,
Copyright © 2007 Inderscience Enterprises Ltd.
W. Song, W. Zhuang and A. Saleh
wireless Internet application projects. He has also developed Bell Canada technology strategy
for the IEEE 802.11/WiFi, IEEE 802.16/WiMAX, and the integration of 3G cellular networks
and WLAN. He is also an Adjunct Professor of Computer and Electrical Engineering at the
University of Waterloo, Canada. He is a Fellow of IEE and a Senior Member of IEEE. His
current research interests are in the areas of next-generation wireless networks and wireless
In the last decade, there has been successful deployment
and fast evolution of various wireless networks. Different
generation wireless networks will integrate heterogeneous
technologies to achieve enhanced performance.
attractive and complementary characteristics presented by
cellular networks and Wireless Local Area Networks
(WLANs) make them promising candidates.
Originally aiming at providing high-quality circuit-
switched voice service to mobile users within wide areas,
cellular networks have been well deployed around the
world and have evolved to the Third Generation (3G). Two
the Universal Mobile Telecommunication System (UMTS)
and cdma2000, which are specified by the 3G partnership
projects, that is, 3GPP and 3GPP2, respectively. Both
systems are based on Code-Division Multiple Access
such as Multimedia Message Service (MMS) and Wireless
Application Protocol (WAP) service. In cdma2000, for
example, the nominal 1.25MHz bandwidth can achieve a
demands for bandwidth-intensive data applications.
On the other hand, usually operating at unlicensed
frequency bands, WLANs provide data services with lower
cost. Moreover, the large bandwidth available for WLANs
makes it possible to achieve higher data rates. For example,
a WLAN can have a bandwidth more than 20MHz. IEEE
802.11b operates at the licence-exempt Industrial, Scientific
and Medical (ISM) frequency band from 2.4 to 2.483 GHz.
It extends the physical layer based on Direct Sequence
Spread Spectrum (DSSS) specified in the original 802.11
standard and supports a higher data rate up to 11Mbps.
The subsequent revisions such as 802.11a and 802.11g adopt
Orthogonal Frequency-Division Multiplexing (OFDM) and
offer a maximum rate of 54Mbps at the unlicensed 5 and
extension to the wired ethernet, a WLAN can only cover
a small geographic area. For instance, an 802.11b Access
Point (AP) can communicate with a Mobile Station (MS)
within up to 60m at 11Mbps and up to 100m at 2Mbps
with omnidirectional antennas. Consequently, with lower
cost and much higher data rates, WLANs can effectively
supplement the 3G networks in hotspot areas, where
bandwidth-demanding applications are concentrated. As a
result, by effectively combining 3G cellular networks and
for different application
mobile users can be provided with both ubiquitous
connectivity and high-rate data services in hotspots.
In the following sections, we first discuss the important
challenging issues involved in this integration problem.
Then we briefly review some typical interworking solutions
proposed in the literature. In the concluding remarks, some
open research issues are outlined.
2 Challenges for 3G/WLAN interworking
The heterogeneous technologies employed in 3G cellular
networks and WLANs bring many challenges to the
interworking. Based on different radio access techniques,
in terms of mobility management, security support and
Quality of Service (QoS) provisioning. In order to achieve
seamless integration, these issues should be carefully
addressed while developing the interworking schemes.
After three-generation evolution, relatively mature and
complete technologies have been established in cellular
networks to address issues such as mobility, security,
QoS, etc. With widely deployed infrastructure from radio
access networks to core networks, ubiquitous connectivity is
provided to mobile uses over wide areas. Different mobility
levels are supported from fast vehicles moving on highways
to stationary users in an indoor environment.
In contrast, the WLAN specifications only focus on the
physical layer and Medium Access Control (MAC) layer.
As for the upper layers, it assumes to adopt the same
protocols as those in wired networks, for example, the
links to avoid performance degradation. A de facto WLAN
system is given in Ahmavaara et al. (2003), in which the
layer-2 distribution system connects multiple APs, while
access routers in turn connect the layer-2 distribution
system to an IP backbone network. In some proposed
interworking schemes (to be discussed), the access routers
offer rich functionalities more than the basic function of
IP routing, for example, transferring authentication and
Through border gateways in the IP backbone network,
WLAN terminals are provided IP connectivity to external
IP networks such as the public Internet or a corporation
intranet. Instead of providing continuous coverage over wide
areas, WLANs are usually disjointly deployed in public or
private hotspots such as cafés, airports and offices. Users
in these areas normally have a very low mobility level, as
most of these areas are located in indoor environments.Also,
cellular coverage is available in these areas. As a result, a
non-uniform overlay topology structure has to be considered
for 3G/WLAN integration.
Interworking of 3G cellular networks and wireless LANs
2.1 Seamless roaming across 3G cellular
networks and WLANs
Taking into account the distinct mobility management
a rather challenging task to support seamless roaming across
the two networks. Either the cellular networks or WLANs
should have inherent mechanisms for location and handoff
management to support the layer-2 or link layer mobility.
In 3G networks (e.g., UMTS or cdma2000), with the aid
of core networks, tunnelling protocols are used to support
roaming within the Public Land Mobile Network (PLMN)
or across 3G PLMNs of different operators with roaming
agreements. For example, in UMTS, the General Packet
Radio Service (GPRS) Mobility Management (GMM) is
used. However, link layer mobility is not enough to provide
network-layer transparency to upper-layer applications.
To avoid disruption of upper-layer sessions due to IP address
changes when user mobility results in changes of network
attachment points, the network-layer IP mobility is needed.
In cdma2000, Mobile IP (MIP) (Perkins, 2002) provides
IP mobility within the same Packet Data Serving Node
(PDSN) and between different PDSNs. However, in current
a UMTS network with the same Gateway GPRS Support
Node (GGSN). It is being considered to introduce IP for
inter-UMTS or inter-technology IP mobility through a three-
state evolution in 3GPP (2000).An overview of the mobility
management in UMTS and cdma2000 can be found in Pang
et al. (2004).
The mobility management in WLANs is much simpler
since they are only oriented to local areas. In IEEE
802.11WLANs, the distribution system (e.g., an 802.3-type
Extended Service Set (ESS). Each BSS is under the control
of an AP in the infrastructure mode. In this case, mobility
The Inter-Access Point Protocol (IAPP) specified in 802.11f
further facilitates the user roaming betweenAPs of different
vendors. When IP connectivity is provided in the WLAN
system, IP micromobility protocols can be introduced to
further support IP layer mobility.
and WLANs in mobility management, to achieve seamless
roaming across the two networks, either a unified mobility
management mechanism is followed, or both networks
maintain proper interoperation.
2.2 Enhanced security level
Network security covers diverse issues such as user
authentication, data confidentiality and integrity and key
management. The security mechanisms in 3G networks such
as UMTS are built upon those used in the second-generation
communications (GSM). In particular, user authentication
in UMTS adopts the Authentication and Key Agreement
(AKA) procedure, which relies on the Universal Subscriber
Identity Module (USIM) running in a smart-card at the
user terminal. In addition to authenticating the subscriber’s
identity, cipher and integrity session keys are also generated
from the long-term preshared secret key stored in the
USIM module and Home Location Register/Authentication
Centre (HLR/AuC). A good introduction to the access
security in 3G networks is given in Koien (2004); Rose and
In the original 802.11 standard, rather weak security is
provided due to lack of key management and flaws of the
it fails the claimed objectives with respect to user identity
privacy and data confidentiality. The 802.11i standard aims
control standard IEEE 802.1X is introduced to enhance
authentication and key management using the Extensible
Authentication Protocol (EAP) (Aboba et al., 2004).Also, as
an interim solution, Temporal Key Integrity Protocol (TKIP)
is specified to fix the vulnerabilities of WEP, while the
long-term solution is based on Advanced Encryption
Standard (AES) in place of the stream cipher RC4 used in
WEP. More details on 802.11 security are given in Edney
For 3G/WLAN interworking, if comparable security cannot
be provided by both networks, adversaries can break into
the system through the weakest component in the security
level. Also, appropriate independence between them should
effect when one of them is broken.
2.3 Consistent end-to-end QoS guarantee
The shared nature of radio link necessitates proper MAC
to coordinate multiple connections to access the shared
multiple access. Moreover, by introducing packet-switched
mode a higher resource utilisation can be achieved for bursty
data traffic by statistical multiplexing. For the multiple
access uplink (from MS to base station), a two-phase
request-grant access procedure is used in 3G networks.
First MSs send transmission requests to the Base Station
(BS) through a contention channel. The BS acknowledges
those successful requests and reserves resources for data
transmission to follow. Then the MSs are notified the
resource assignments. This type of centralised control and
reservation-based resource allocation, together with proper
admission control to limit the traffic load, enables fine QoS
provisioning in 3G cellular networks.
On the other hand, the original WLAN specifications
only support best-effort data service with contention-based
random access protocol, for example, the Distributed
Coordination Function (DCF) of 802.11 WLAN based
on Carrier Sense Multiple Access/Collision Avoidance
(CSMA/CA). In this mode, the AP competes for access
with MSs instead of scheduling the resource assignments
as a BS in 3G networks. The other centralised control
mode, Point Coordination Function (PCF), is based on
polling by the AP. It is rarely implemented in reality due to
unresolved problems such as uncontrolled transmission time
of polled MSs. The distributed and contention-based control
leads to weak QoS support capability. To achieve better
W. Song, W. Zhuang and A. Saleh
QoS, 802.11e develops new approaches by means of MAC
enhancements. The above two access modes are improved
with service differentiation. Also, admission control and
multimedia applications with stringent QoS requirements.
Various QoS mechanisms for 802.11 are explored in Zhu
et al. (2004). Nevertheless, WLANs still cannot be expected
to support the same level of QoS as 3G networks.
Considering the differences of 3G networks and WLANs
in QoS provisioning and the aforementioned overlaying
structure, different services can be admitted to either
the cellular network or the WLAN according to their
traffic characteristics and QoS requirements. For example,
whereas the delay-tolerant data traffic can be admitted to the
WLAN to enjoy the high throughput. It is also an important
issue to maintain consistent or smoothly adapted QoS during
vertical handoff (i.e., handoff between the cellular network
and the WLAN).
2.4 Interworking scenarios
Currently, WLANs may be owned by a 3G operator, a
(e.g., airports or property management corporations) or
a business enterprise for internal use (Salkintzis, 2004).
The interworking mechanisms are directly related to the
ownership of WLANs. More exposure is possible to
integrate the WLANs owned by the cellular operator itself.
The objective penetration level between the 3G networks
and WLANs and the provisioned services lead to different
requirements for the interworking mechanisms. The services
supported in future integrated 3G/WLAN networks are
envisioned in Axiotis et al. (2004). However, the integration
of the two technologies will be a gradual development
process. From the perspective of a 3G operator,
step-wise approach is proposed in 3GPP (2003b), which
defines six interworking scenarios with each scenario
specifying an incremental set of service and operational
features. The first scenario only requires common billing
and customer care. In the second scenario, a 3G subscriber
roaming to the WLAN is authenticated and charged by its
3G home network. Only IP access service via the WLAN
is provided to the roaming user. In the third scenario, the
3G packet-switched services are also open to users attached
to the WLAN, such as MMS, WAP service, IP multimedia
and location-based services. The fourth and fifth scenarios
improve the third scenario with higher requirement for
service continuity. The sixth scenario allows the access to
the 3G circuit-switched services (such as conventional voice
andWLANs. So far, a lot of research in the literature focuses
on the four scenarios in the middle.
3 3G/WLAN interworking architectures
The standardisation for 3G/WLAN interworking is well in
progress by 3GPP and 3GPP2. The high-level interworking
requirements, architecture and procedures (e.g., network
specified in 3GPP (2004a). In ETSI (2001) the integration
architecture is classified into two categories according to the
interdependence between the two access networks, that is,
tight coupling and loose coupling. Different interworking
mechanisms are developed to support mobility, security,
QoS, charging and billing in the tightly coupled and loosely
coupled heterogeneous networks.
etc.) have been
3.1 Tightly coupled and loosely
In the tight-coupling architecture, the WLAN is connected
to the 3G core network as one 3G radio access network.
Figure 1 illustrates a simplified architecture for interworking
UMTS/GPRS networks with 802.11 WLANs. The lines
with tags ‘a’ and ‘b’ are examples for tight coupling
with different integration points. Typical tight-coupling
architectures include the first architecture proposed in
Salkintzis et al. (2004) and Buddhikot et al. (2003b).
We can see that, in this type of integration architecture,
the cellular radio is simply replaced with WLAN radio
providing equivalent functions. As a consequence, the 3G
protocols and existing network infrastructures can be reused.
For example, the user roaming across the two domains is
thus enhancing the interdomain mobility management
for highly mobile users in hostile outdoor environments may
not operate properly for WLANs (Pichna et al., 2000). The
main disadvantages of the tight coupling approach include:
1 an interface in 3G core networks exposed to WLAN is
required, which is a challenge as the two domains are
likely developed and deployed independently by
2 a large volume of WLAN traffic will go through the 3G
core network, possibly making the latter a network
3 and the WLAN needs to have a protocol stack
compatible with that of the 3G networks.
On the contrary, for the loose-coupling approach (shown
with the line ‘c’ in Figure 1), the WLAN is connected
to the cellular core network indirectly through an external
IP network such as the Internet. The second architecture
proposed in Salkintzis et al. (2004) and Buddhikot et al.
(2003b), respectively, and the operator WLAN system
in Ala-Laurila et al. (2001) belong to this category.
This type of architecture imposes minimal requirements
to modify current WLAN standards, and allows for the
flexibility and independence of implementing individually
different mechanisms within each network. However, the 3G
such as MIP for mobility management and AAA support.
Moreover, as the two domains are separated, the mobility
signalling may traverse a relatively long path, thus inducing
enhancement mechanisms for MIP to reduce the handoff
latency, such as regional registration and dynamic Home
Interworking of 3G cellular networks and wireless LANs
Interworking architecture for 3GPP UMTS and IEEE 802.11 WLANs: (a) WLAN integrated at SGSN; (b) WLAN
integrated at GGSN and (c) WLAN integrated to external IP networks
Agent (HA) assignment. Overall, the loose coupling is the
it allows the gradual deployment of hotspots with no or little
modification on the 3G networks (Buddhikot et al., 2003a,b;
Koien and Haslestad, 2003).
It is well recognised that the future networks will be
all-IP networks. Therefore, it is a natural choice to glue
3G networks and WLANs with the pervasive IP technology.
In fact, we can see from the later sections that many
mechanisms proposed for the interworking follow the de
facto standards in the Internet community.
3.2 Authentication and authorisation support
It is expected that in a 3G/WLAN integrated network the
wireless terminal will be dual-mode, which means that the
terminal will be equipped with network interfaces to both 3G
networks and WLANs (Axiotis et al., 2004). However, only
one subscription is needed with a 3G operator or a WLAN
service provider, who has roaming agreements to support the
interworking (Buddhikot et al., 2003a). A lot of research in
the literature considers the scenario in which a 3G subscriber
is provided WLAN access via independent WLAN systems.
Take the interworking of UMTS and 802.11WLAN as
an example. As mentioned in Section 2, 802.11WLANs
adopt 802.1X for access control, based on the EAP and
AAA framework. EAP sets no restriction on specific
authentication methods, while the AAA protocols, such
as remote authentication dial-in user service (RADIUS)
or its enhanced version Diameter,
attributes in the authentication messages. The selection of
RADIUS as the AAA protocol is not mandatory in the
802.1X or 802.11i; but it is in Wi-Fi protected access
(WPA) for interoperability. Considering the flexibility and
interoperability requirement, AAA server is introduced in
UMTS (3GPP, 2004b). By means of EAP-AKA, the 3G
home AAA server authenticates the UMTS subscribers
roaming to a WLAN through the UMTS AKA procedure
(Ahmavaara et al., 2003; Buddhikot et al., 2003a; Koien
and Haslestad, 2003; Salkintzis, 2004), while EAP-SIM can
be used for legacy GSM/GPRS system (Ala-Laurila et al.,
2001; Salkintzisetal., 2002).Theauthenticationinformation
and subscription profile can be retrieved from the Home
Subscriber Server (HSS) or HLR. Similar arrangements are
implemented in the cdma2000/WLAN interworking.
There are three entities defined in 802.11X, that is,
supplicant, authenticator and authentication server. It is
natural that the wireless terminal and the 3G AAA server
will perform the functions of the supplicant and the
authentication server, respectively. The authenticator is
responsible for relaying EAP frames sent by the supplicant
via EAP-Over-LAN (EAPOL), and repacketing them into
appropriate AAA messages for onward transmission to the
authentication server, and vice versa. In the interworking
mechanism, the functions of the authenticator can be carried
by the AP (Salkintzis et al., 2004), the WLAN access
router, or a separate WLAN AAA proxy (Salkintzis, 2004).
A successful authentication procedure is summarised in
In the operator WLAN architecture proposed for legacy
GSM/GPRS networks in Ala-Laurila et al. (2001), the
access controller actually acts as the access router shown in
Figure 1. It provides IP routing for WLANs and also relays
authentication messages for the terminal and authentication
and the authentication server is based on RADIUS, and
the SIM-based authentication is executed with the aid of
HLR. However, the authentication signalling between the
mobile terminal and the access controller is the operator
WLAN-specificNetworkAccess Authentication and
W. Song, W. Zhuang and A. Saleh
Authentication procedure based on EAP-AKA via homeAAA server
Accounting Protocol (NAAP), which encapsulates GSM
authentication messages inside IP packets. This actually
limits the applicability of the proposed interworking
mechanism and is not compatible with current WLAN or
at layer 2. When MIP is supported in the 3G and WLAN
networks, the authentication can also be carried out at
the network (IP) layer (Buddhikot et al., 2003a). In the
mobile IP mode of cdma2000, PDSN is augmented with
Foreign Agent (FA) functionality, while HA is introduced
to maintain location information of MSs and forward IP
packets to MSs via tunnelling. For UMTS, the technical
report 23.923 (not technical specification yet) (3GPP,
2000) outlines the plan to implement MIP in UMTS IP
core network through three steps. In the first stage, the
FA functionality is added to GGSN, which can support
roaming between WLANs and UMTS via MIP. Stage
II provides IP mobility for inter-GGSN handoff, while
in stage III GGSN and SGSN are further combined to
one Intelligent GPRS support node (IGSN) similar to
the PDSN in cdma2000. With MIP being implemented,
the MS roaming to a WLAN can be authenticated with
the MIP registration procedure. An authentication example
for cdma2000/WLAN interworking is illustrated in Figure 3.
After obtaining a new Care-of Address (CoA) from the
Agent advertisements of FA, the MS initiates the
request message to the FA, which will contact the
3G home AAA RADIUS server to authenticate the
roaming user via mobile-foreign challenge extension and
mobile-AAA authentication extension (Perkins et al., 2004).
In the mobile IP mode of cdma2000, neither Challenge
Handshake Authentication Protocol (CHAP) nor Password
Authentication Protocol (PAP) is performed as in the
simple IP mode, because an additional AAA traversal
will result in longer initial setup time and reestablishment
time (3GPP2, 2002). After successful authentication by
theRADIUSserver, the MIP registration request
of the MS will be sent to the HA via HA request
message. After receiving HA response indicating the
registration is validated by the HA, the AAA RADIUS
server acknowledges the PDSN/FA with RADIUS access
accept. The PDSN/FA will subsequently send an MIP
registration reply message to the MS terminating
the authentication and registration procedure.
In addition, it is possible to perform the authentication
at the application layer. In Buddhikot et al. (2003a), it is
proposed to authenticate the mobile node via a secured web
page for login over a Hypertext Transport Protocol Secured
(HTTPS) connection. This type of solution has restricted
applications since it requires the support of specific services
such as HTTP.
In summary, by introducing the flexibleAAA framework,
the 3G-specific authentication mechanisms can be reused.
At the same time, independent WLAN service providers
can implement their own preferred authentication methods,
which aregenerally the
Internet community, such as Extensible Authentication
Protocol-Message Digest number 5 (EAP-MD5) and
authentication and authorisation procedure with mobility
management, the signalling procedure can be simplified.
However, the authentication and authorisation procedure
should be efficient enough to minimise its impact
on handoff latency. As for higher-layer authentication
mechanisms, although more freedom is allowed for
underlying technologies, the messaging delay and specific
application restriction are their main drawbacks.
popular standards in the
hand, by couplingthe
Interworking of 3G cellular networks and wireless LANs
IP-level authentication procedure based on MIP registration
3.3 Mobility management across 3G networks
Mobility management consists of two aspects, that is,
location management and handoff management. Location
handoff management maintains the ongoing connections
while switching attachment points. The mobility protocols
and application layers (Banerjee et al., 2003). Link-layer
for example, the GMM in UMTS/GPRS networks. In the
tightly coupled interworking architecture, the WLANs are
integrated as any other 3G radio access networks. As a
result, the 3G mobility management mechanism can be
reused to support mobility across WLANs and 3G networks
(Buddhikot et al., 2003b; Salkintzis et al., 2002).
The most popular network layer mobility protocol is
Mobile IP (Perkins, 2002). By introducing mobility agents
and IP tunnelling, the upper layer applications are enabled
transparency to IP address changes due to user movement.
However, the original MIP protocol suffers from the triangle
routing problem. The packets have to be first routed to the
HA before being tunnelled to the mobile node, while a
mobile node sends its packets through a router on the foreign
network assuming the routing is independent of source
address. When this assumption is not valid, topologically
correct reverse tunnels can also be established from the
CoA to the HA. As a consequence, an extra delay results
from the long route when the visited foreign network is far
from the home network.This problem can be solved by route
optimisation, in which a direct route is established between
the mobile node and its correspondent node. In addition,
the location registration to the HA may lead to a heavy
signalling load when there are a large number of mobile
networks are far apart. Therefore, MIP is more suited for
macromobility with infrequent movement and often between
different administrative domains (interdomain). There are
many MIP variants to support micromobility, which refers
to mobility within one domain (intradomain). The handoff
latency can be reduced by means of localising signalling
via regional/hierarchical registration (tunneling-based) or
For tunnel-based micromobility Protocols, there are Mobile
IP regional registration, hierarchical MIP and Intra-Domain
Mobility Management Protocol (IDMP). Typical routing-
based micromobility protocols include cellular IP and
HAWAII (Akyildiz et al. 2004).
in Salkintzis et al. (2004) and Buddhikot et al. (2003a), the
MIP approach is used for the mobility across the cellular
network and WLANs. Even though the loose coupling
mobility management schemes within each network, it
has become a trend to apply MIP in 3G networks to
pave the way for future all-IP networks. A cross-layer
solution is proposed in Akyildiz et al.
support mobility in all-IP-based wireless networks with
heterogeneous access technologies. The architecture can be
applied to the 3G/WLAN interworking case. In the scheme,
the aforementioned micromobility protocols are used for
intradomain mobility, while a cross-layer mobility protocol
is proposed for interdomain mobility.With early detection of
possible interdomain handoff using link layer information,
it is enabled to carry out authentication, authorisation, and
MIP registration before the actual handoff. As a result, the
interdomain handoff delay is reduced to be comparable to an
intradomain handoff delay.
Moreover, transport layer mobility protocols can be used
to prevent transport-layer applications such as Transport
Control Protocol (TCP) connections from being interrupted
by IP address changes due to user mobility. One solution
in this type of mobility support is TCP-Migrate (Snoeren
W. Song, W. Zhuang and A. Saleh
and Balakrishnan, 2000). Another transport layer scheme
proposed in Ma et al. (2004) supports UMTS/WLAN
vertical handoff via Stream Control Transmission Protocol
(SCTP). Although performing mobility management at the
transport layer provides network-independence, it requires
more functions carried by end systems and also many
modifications to current end systems.
mobility support independent of underlying wireless access
technologies and network layer protocols (Ma et al.,
2004). Hence, they offer another alternative for mobility
management in the 3G/WLAN interworked networks.
A typical representative of application layer mobility
2003), which supports pre-call and mid-call mobility with
the aid of application-specific redirect server. Although the
application layer solution introduces less modification to
existing protocol stacks and infrastructures of 3G networks
and WLANs, longer handoff latency may be incurred due
to the high-layer messaging. In Politis et al. (2004), a
hybrid multilayer mobility management scheme is proposed.
To handle macromobility, it uses MIP for non-real time
services and SIP for real-time services. Existing protocols
such as cellular IP, HAWAII and hierarchical MIP are
employed to support micromobility. AAA context transfer
is further introduced to avoid additional delay induced by the
AAA security procedure during handoff.
In summary, for mobility management in 3G/WLAN
integrated networks, accurate location information of MSs
by paging, so that the MSs can move freely across the two
domains. Moreover, the vertical handoff is required to be
fast, smooth and seamless (Gao et al., 2004). Especially for
real-time services, the vertical handoff latency needs
to satisfy strict delay bounds. Smooth handoff requires
minimised packet loss during handoff, while the perceptible
service interruption should be minimised to achieve
the objective of seamless handoff.
interworking environment involved with multiple domains,
when enhanced with micromobility protocols, the Mobile
IP approach is a good solution and can well meet
the above requirements, especially when a hierarchical
layered mobility management architecture is applied.
The performance of MIP can be further improved with
cross-layer design. Architecturally, the network layer is
the right place to handle mobility (Banerjee et al., 2003).
The higher-layer approaches such as the transport and
application-layer mobility management schemes have the
advantage of retaining the network interfaces unchanged.
This feature applies well to the heterogeneous nature of
3G/WLAN integrated networks. Nonetheless, compared
with the network-layer approaches,
solutions are relatively immature and incomplete.
In a 3G/WLAN
3.4 QoS provisioning architecture
The next-generation wireless networks are required to
for the 3G/WLAN interworking. Multimedia applications
usually feature more bandwidth-demanding and stringent
QoS requirements. We have mentioned in Section 2.3
the differences between 3G networks and WLANs in
QoS support at the physical layer and MAC layer. This
heterogeneous nature results in technical challenges in
end-to-end QoS provisioning for a variety of services in an
integrated 3G/WLAN network.
Figure 4 demonstrates the various scenarios requesting
end-to-end services. It can be seen that different networks
may be involved along the end-to-end path, for example,
3G radio access networks, 3G core networks, WLAN access
networks, IPbackboneandthewiredInternet. Insuchacase,
the following aspects should be well addressed in the QoS
provisioning architecture (Fodor et al., 2003):
QoS attributes – The traffic characteristics and
end-to-end QoS requirements need to be clearly
specified by the QoS attributes. These QoS attributes
should be independent of the underlying access
technologies, easily interpreted and mapped to service
requirements for the networks involved along the
QoS signalling protocols – The QoS signalling
protocols are responsible for QoS negotiation, resource
reservation, etc.Although they are not required to be
identical across different network domains, it is
essential for the QoS attributes to be distributed via
signalling along the end-to-end path without
modification and being kept consistent.
QoS mechanisms – To satisfy the QoS requirements
presented to each network segment along the
end-to-end path, proper local QoS mechanisms such as
traffic control and scheduling are evoked to provide the
requested QoS. Distinct QoS mechanisms are employed
in 3G networks and WLANs, which results in different
levels of QoS support capability.
in Dixit et al. (2001). For UMTS, a layered QoS architecture
is specified (3GPP, 2003a). The end-to-end bear required to
meet the QoS objectives is decomposed into a concatenation
of bear services, which are provided by the different network
segments involved. There are four traffic classes supported
in UMTS, that is, conversational, streaming, interactive and
background, with QoS attributes defined for each class, such
as maximum bit rate, transfer delay and traffic handling
priority. As for the QoS signalling, the UMTS Packet Data
Protocol (PDP) context signalling mechanism is used for the
bear services within UMTS. Diverse QoS schemes can be
employed in the air interface, radio access networks and core
networks to achieve the QoS objectives, for example, power
control, bandwidth scheduling and transport technologies.
As mentioned in Section 3, IP technologies offer a
common platform to integrate WLANs with 3G networks.
In UMTS, the gateway node to external packet networks
(i.e., GGSN) is required to provision an IP bearer service
manager for interworking with external IP networks. For
IP networks,there are two major QoS architectures,
that is, Integrated Services (IntServ) and Differentiated
Services (DiffServ). In the IntServ architecture, with the
aid of Resource Reservation Protocol (RSVP), resources
are reserved along the end-to-end path to satisfy the QoS
Interworking of 3G cellular networks and wireless LANs
End-to-end service in 3G/WLAN interworked networks
requirements for each flow. However, the per-flow soft states
must be maintained end-to-end and refreshed periodically
to hold the reservation. As a result, this model is not well
scalable for large networks with a large population.
On the other hand, DiffServ does not need a separate
signalling protocol and it only exerts control for aggregate
traffic. At the network edge, the traffic flows are classified
into one of several classes according to the Service Level
and between different network domains (Grilo et al.,
2003). The Type Of Service (TOS) field of IP packets is
marked with the Differentiated Service Code Point (DSCP)
indicating the service class. Then the packets will receive
priority-based treatment according to Per-Hop Behaviour
(PHB) defined for each service class. PHBs are actually
implemented by means of buffer management and packet
scheduling mechanisms (Moon and Aghvami, 2003). In this
way, service differentiation can be provided to different
classes. In addition, the edge nodes can be augmented
with functionalities such as traffic policing and admission
control to prevent unacceptable QoS degradation due to
saturation. As such, DiffServ only provides relative QoS
to aggregate traffic flows instead of absolute QoS guarantee
to individual flows as IntServ. Hence, although DiffServ has
good scalability, only coarse QoS provisioning is enabled.
Because both IntServ and DiffServ models are proposed
for wired IP networks, unique characteristics of wireless
links and user mobility should be taken into account for
extending them to wireless/mobile networks. The extension
mechanisms of RSVP and DiffServ for IP-based wireless
mobile networks are surveyed in Moon andAghvami (2001)
and Moon and Aghvami (2003), respectively. Especially,
in the case of integrated 3G and WLANs networks, when
vertical handoff occurs between the 3G networks and
WLANs, the bear services offered by the two access
technologies provide different QoS support. These changes
should be considered in the QoS context transfer and
service reestablishment after handoff.
In Manner et al. (2002), an IP QoS model combining
IntServ and DiffServ is proposed for heterogeneous wireless
networks. In the wireless access networks,
architecture is applied, since the traffic volume is relatively
low and scalability will not be a severe problem. Also,
the scarce wireless bandwidth can be efficiently utilised by
properly allocating resources to each traffic flow. On the
for scalability purpose. At the edge of the sender’s access
networks, the RSVP signalling for end-to-end resource
reservation is mapped to DiffServ PHBs, and the RSVP
messages are tunnelled through the DiffServ domains till
the receiver end. The DiffServ approach for core networks
is further justified by the bandwidth over-provisioning
in current core networks and the relatively low cost of
upgrading transmission capability (Marques et al., 2003).
A similar idea is used in the QoS architecture proposed
in Grilo et al. (2003), where RSVP is slightly modified.
Instead of exchanging the RSVP PATH and RSVP RESV
messages end-to-end from the sender to the receiver, the
RSVP signalling is localised within the access networks
via proxies. This facilitates the reservation path repair after
handoff. In addition, it is proposed inAlam et al. (2001) that
the DiffServ domain. Due to the intrinsic QoS provisioning
capability, MPLS can facilitate scalability by means of flow
aggregation and at the same time guarantee individual QoS
without the need to maintain per-flow awareness along the
path (Alam et al., 2001).
To achieve better QoS support, mobility management and
QoS provisioning can be coupled in the architecture design.
In Lo et al. (2004), a mobility and QoS architecture is
proposed for all-IP based 4G wireless networks. It takes
advantage of the two-layer hierarchical structure of IDMP
W. Song, W. Zhuang and A. Saleh
mobility management to reduce the resource reservation
delay due to RSVP reestablishment after handoff. The
per-flow RSVP reservation is terminated at a fixed anchor
subnet agent, while a forwarding chain is used to track the
host mobility within a domain. As such, the reservation
messages can be effectively restricted within a local scope.
Another IP-based QoS architecture is proposed in
Marques et al. (2003) for 4G networks supporting multiple
accesses and multiple service providers. In this architecture,
the end-to-end QoS support is integrated with mobility
and Authentication, Authorisation, Auditing and Charging
(AAAC). In each network domain, anAAAC system is used
to handle the network access control, while at least one
QoS Broker (QoSB) manages the access network resources.
The Common Open Policy Service (COPS) protocol is used
for information exchanging between QoSBs, edge routers
andAAAC systems, because in this architecture QoSBs and
AAAC systems act as Policy Decision Points (PDPs) while
edge routers act as Policy Execution Points (PEPs).
In Zhuang et al. (2003), a policy-based multidomain
QoS architecture is proposed to provide consistent QoS
control over an integrated UMTS and WLAN system.
The architecture varies with different interworking scenarios
such as interworking with a UMTS operator’s WLAN, a
WLAN shared by multiple operators or aWLAN of a UMTS
operator’s customer. This policy-based type of approach
views the development of proper QoS architecture for
3G/WLAN interworking from a new perspective. However,
there are still some problems that need more careful
considerations, such as fastening policy negotiation and
securing information exchanging between interconnected
policy entities (Zhuang et al., 2003).
In this paper, we have introduced the complementary
characteristics of 3G networks and WLANs and the
wireless technologies lead to many challenging issues that
roaming, enhanced security and consistent QoS, there
are a variety of interworking mechanisms proposed from
different perspectives. In particular, we examine the typical
solutions for user authentication, mobility management and
QoS provisioning in the 3G/WLAN interworked networks.
It is observed that IP technologies play an important
role in gluing the two heterogeneous technologies. The
demand for ubiquitous Internet access further strengthens
this convergence trend. For instance, the well-known AAA
framework, MIP protocol, IntServ/RSVP and DiffServ QoS
architectures are considered as promising approaches for
interworking 3G networks and WLANs.
the local QoS mechanisms implemented in the 3G
networks and WLANs can provide certain levels of QoS
support, proper resource allocation schemes are needed to
utilise the integrated resources effectively and efficiently.
In Song et al. (2005), we investigate the resource allocation
problem in 3G/WLAN integrated networks. In the proposed
resource allocation scheme, we try to take advantage
of the characteristics of this heterogeneous network with
multiple services, so that better QoS assurance and higher
utilisation can be achieved. In Luo et al. (2003), a radio
resource scheduling scheme is proposed for a 3G/WLAN
coupled network. Based on the traffic characteristics, QoS
and splits the traffic over the two networks. A certain
amount of resources in each network is allocated to the
corresponding traffic substream to achieve the target QoS.
A higher utilisation is available due to a larger trunking gain
and effective manipulation of the overall resources of the
two networks. However, these are only some initial research
an open issue that needs more in-depth investigation.
Aboba, B., Blunk, L., Vollbrecht, J., Carlson, J. and Levkowetz, H.
(Eds) (2004) ‘Extensible authentication protocol (EAP)’, IETF
Ahmavaara, K., Haverinen, H. and Pichna, R. (2003) ‘Interworking
architecture between 3GPP and WLAN systems’,
Communications Magazine, Vol. 41, No. 11, pp.74–81.
Akyildiz, I.F., Xie, J. and Mohanty, S. (2004) ‘A survey of mobility
management in next-generation all-IP-based wireless systems’,
IEEE Wireless Communications, Vol. 11, No. 4, pp.16–28.
Ala-Laurila, J., Mikkonen, J. and Rinnemaa, J. (2001) ‘Wireless
LAN access network architecture for mobile operators’, IEEE
Communications Magazine, Vol. 39, No. 11, pp.82–89.
Alam, M., Prasad, R. and Farserotu, J.R. (2001) ‘Quality
of service among IP-based heterogeneous networks’, IEEE
Personal Communications Magazine, Vol. 8, No. 6, pp.18–24.
Axiotis, D.I., Al-Gizawi, T., Protonotarios, E.N., Lazarakis, F.I.,
Papadias, C. and Philippopoulos, P. (2004) ‘Services in
interworking 3G and WLAN environments’, IEEE Wireless
Communications, Vol. 11, No. 5, pp.14–20.
Banerjee, N., Wu, W. and Das, S.K. (2003) ‘Mobility support
in wireless Internet’, IEEE Wireless Communications, Vol. 10,
No. 5, pp.54–61.
Buddhikot, M.M., Chandranmenon, G., Han, S., Lee, Y-W.,
Miller, S. and Salgarelli, L. (2003a) ‘Design and implementation
of a WLAN/cdma2000 interworking architecture’,
Communications Magazine, Vol. 41, No. 11, pp.90–100.
Buddhikot, M.M., Chandranmenon, G., Han, S., Lee, Y.W.,
Miller, S. and Salgarelli, L. (2003b) ‘Integration of 802.11 and
third-generation wireless data networks’, Proceedings of the
IEEE INFOCOM’03, Vol. 1, pp.503–512.
Campbell, A.T., Gomez, J., Kim, S., Wan, C-Y., Turanyi, Z.R. and
Valko,A.G. (2002) ‘Comparison of IP micromobility protocols’,
IEEE Wireless Communications, Vol. 9, No. 1, pp.72–82.
Dixit, S., Guo,Y. and Antoniou, Z. (2001) ‘Resource management
and quality of service in third generation wireless networks’,
IEEE Communications Magazine, Vol. 39, No. 2, pp.125–133.
Edney, J. and Arbaugh, W. (2003) Real 802.11 Security: Wi-Fi
Protected Access and 802.11i,Addison Wesley Professional.
ETSI (2001) ‘Requirements and architectures for internetworking
between HIPERLAN/2 and 3rd generation cellular systems’,
ETSI TR 101 957.
Interworking of 3G cellular networks and wireless LANs Download full-text
Fodor, G., Eriksson, A. and Tuoriniemi, A. (2003) ‘Providing
quality of service in always best connected networks’, IEEE
Communications Magazine, Vol. 41, No. 7, pp.154–163.
Gao, X.,Wu, G. and Miki,
provisioning in mobile heterogeneous networks’, IEEE Wireless
Communications, Vol. 11, No. 3, pp.24–34.
3GPP (2000) ‘Combined GSM and Mobile IP mobility handling in
UMTS IP CN’, 3GPP TR 23.923 V.3.0.0.
3GPP2 (2002) ‘Wireless IP network standard’, 3GPP2 P.S0001-B
3GPP (2003a) ‘Quality of service (QoS) concept and architecture’,
3GPP TS 23.107 V5.12.0.
3GPP (2003b) ‘Feasibility study on 3GPP system to wireless local
area network (WLAN) interworking’, 3GPP TR 22.934 V6.2.0.
3GPP (2004a) ‘3GPP system to wireless local area network
(WLAN) interworking; system description (release 6)’, 3GPP
TS 23.234 V6.3.0.
3GPP (2004b) ‘Wireless local area network (WLAN) interworking
security (release 6)’, 3GPP TS 33.234 V6.3.0.
Grilo, A., Nunes, M., Sergio, G. and Ciulli, N. (2003) ‘Integration
of IP mobility and QoS for heterogeneous wireless access in
MOICANE’, Proceedings of the IEEE Symposium on Personal,
Indoor and Mobile Radio Communications (PIMRC’03),Vol. 1,
Koien, G.M. (2004) ‘An introduction to access security in UMTS’,
IEEE Wireless Communications, Vol. 11, No. 1, pp.8–18.
Koien, G.M. and Haslestad, T. (2003) ‘Security aspects of
3G-WLAN interworking’, IEEE Communications Magazine,
Vol. 41, No. 11, pp.82–88.
for mobility and QoS support in all-IP wireless networks’, IEEE
Journal of Selected Areas on Communications, Vol. 22, No. 4,
Luo, J., Mukerjee, R., Dillinger, M., Mohyeldin, E. and Schulz, E.
(2003) ‘Investigation of radio resource scheduling in WLANs
coupled with 3G cellular network’, IEEE Communications
Magazine, Vol. 41, No. 6, pp.108–115.
Manner, J., Burness, L., Hepworth, E., Lopez, A., and Mitjana, E.
(2002) ‘Provision of QoS in heterogeneous wireless IP access
networks’, Proceedings of the IEEE Symposium on Personal,
Indoor and Mobile Radio Communications (PIMRC’02),Vol. 2,
Marques, V., Aguiar, R.L., Garcia, C., Moreno, J.I., Beaujean, C.,
Melin, E. and Liebsch, M. (2003) ‘An IP-based QoS architecture
for 4G operator scenarios’, IEEE Wireless Communications,
Vol. 10, No. 3, pp.54–62.
Ma, L., Yu, F., Leung, V.C.M. and Randhawa, T. (2004)
‘A new method to support UMTS/WLAN vertical handover
using SCTP’, IEEE Wireless Communications, Vol. 11, No. 4,
T. (2004) ‘End-to-end QoS
Moon, B and Aghvami, H. (2001) ‘RSVP extensions for real-time
services in wireless mobile networks’, IEEE Communications
Magazine, Vol. 39, No. 12, pp.52–59.
Moon, B., and Aghvami, H. (2003) ‘DiffServ extensions for
QoS provisioning in IP mobility environments’, IEEE Wireless
Communications, Vol. 10, No. 5, pp.38–44.
Pang, A-C., Chen, J-C., Chen, Y-K. and Agrawal, P. (2004)
‘Mobility and session management: UMTS versus cdma2000’,
IEEE Wireless Communications, Vol. 11, No. 4, pp.30–43.
Perkins, C. (Education) (2002) ‘IP mobility support’, IETF
Perkins, C.E., Calhoun, P.R. and Bharatia, J. (2004) ‘Mobile
IPv4 challenge/response extensions (revised)’, Internet Draft,
Pichna, R., Ojanpera, T., Posti, H. and Karppinen, J. (2000)
‘Wireless Internet – IMT-2000/wireless LAN interworking’,
Journal of Communications and Networks, Vol. 2, No. 1,
Tafazolli, R. and Dagiuklas, T. (2004) ‘Hybrid multilayer
mobility management withAAA context transfer capabilities for
all-IPnetworks’, IEEEWirelessCommunications,Vol.11, No.4,
Rose, G. and Koien, G.M. (2004) ‘Access security in CDMA2000,
including a comparison with UMTS access security’, IEEE
Wireless Communications, Vol. 11, No. 1, pp.19–25.
‘WLAN-GPRS integration for next generation mobile data
networks’, IEEE Wireless Communications, Vol. 9, No. 5,
A.K.,Fors, C. and Pazhyannur,R. (2002)
Salkintzis, A.K. (2004) ‘Interworking techniques and architectures
for WLAN/3G integration toward 4G mobile data networks’,
IEEE Wireless Communications, Vol. 11, No. 3, pp.50–61.
to host mobility’, Proceedings of the ACM/IEEE 6th Annual
Song, W., Jiang, H., Zhuang, W. and Shen, X. (2005) ‘Resource
management for QoS support in cellular/WLAN interworking’,
IEEE Network Magazine, Vol. 19, No. 5, pp.12–28.
Zhuang, W., Gan, Y-S., Loh, K-J., and Chua, K-C. (2003)
‘Policy-based QoS-management architecture in an integrated
UMTS and WLAN environment’,
Magazine, Vol. 41, No. 11, pp.118–125.
Zhu, H., Li, M., Chlamtac, I. and Prabhakaran, B. (2004) ‘A survey
of quality of service in IEEE 802.11 networks’, IEEE Wireless
Communications, Vol. 11, No. 4, pp.6–14.