I provide a comprehensive overview on various traffic offloading solutions:
1. Femtocells, which provides the benefits of scalability, automatic configuration and self-optimization.
2. WiFi, widely available in homes and hotspots.
3. Integrating femto and WiFi together to reap the benefits of both femtocell and WiFi technology.
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
Traffic Offloading Solutions: Femto, WiFi and Integrated Femto-WiFi
1. 1
Traffic Offloading Solutions:
Femtocells and WiFi
Shristi Nhuchhe Pradhan
72385115
Electrical and Computer Engineering
The University of British Columbia, Vancouver, Canada
Abstract
With increasing penetration of smartphones, user’s demand for data services is growing unabated.
The challenge is to realize this high traffic demand and at the same time provide users with improved
speed and consistent coverage. Offloading of traffic from the macro network using technologies like
femtocells and wireless fidelity (WiFi) provide significant reduction in traffic flowing into the macro
cellular’s core network and therefore leading to improvement in network capacity. We discuss the
architecture and offloading methods using femtocells and WiFi. Then, we describe integrated femto-
WiFi (IFW) technology for traffic offloading which helps in filling the loopholes of using femtocells or
WiFi alone. We describe the traffic management and architectural aspects of IFW and further discuss
the challenges and issues faced with the deployment of each of these traffic offloading solutions.
2. 2
I. INTRODUCTION
In the last decade, wireless industry experienced tremendous growth, with 3G being widely
deployed and supporting over a billion of users. With the success of broadband, 3G continues
to expand its footprint and 4G technology up in the forefront, leading to increasing demand
of wireless data around the world. At the same time, smartphones are getting smarter with im-
proved user interfaces, increased number of applications and faster processors, and this increased
penetration of data devices have further contributed to growth of mobile data usage. Meeting
this increasing data traffic demand and solving the bandwidth crunch are the biggest challenges
for the entire wireless industry.
Studies have shown that 70-80% of mobile traffic is generated indoor in homes or offices [1].
Providing excellent mobile services indoor poses another challenge for the network operator,
due to the limitations in indoor signal penetration particularly at higher frequencies [2]. There is
growing demand on macro networks to provide reasonable data during high load periods with the
increase in number of subscribers and data usage. With limited spectrum, it becomes essential to
manage macro network resources intelligently. Cell splitting might be an approach to solve this
problem, where the operator deploys more cell site to ease pressure on the network, but, there
are limits on what the network operators can achieve with the traditional cell splitting, which
still doesn’t address the problem of weak in-building coverage [3].
To address these issues, there has been much interest in a new paradigm called heterogeneous
networks (HetNets), where smalls cells, like femto, pico or micro overlay the macrocell network.
Femtocells are deployed in residences, offices or hotspots for the benefit of traffic offloading
from the macro network, which improve spatial reuse and coverage and allow cellular systems to
achieve high data rates. Alternately, IEEE 802.11 wireless fidelity (WiFi) is another compelling
option, which allows traffic offload from macro network using unlicensed band [4]. Further, these
two technologies can be merged together to form the integrated femto-WiFi (IFW), providing
the benefits of both femtocells and WiFi [5].
We achieve traffic offloading by using alternate network technologies for delivering traffic
originally targeted for cellular networks. In a traditional cellular network, all the traffic to and
from user equipment (UE) travel all the way from the UE to a cell site that might be located
a fraction of a mile away to several miles away. Traffic offloading using femtocells or WiFi
router would carry the traffic from UE to the cell site through an alternate network, typically
3. 3
Fig. 1: Femtocell network architecture
a digital subscriber line (DSL) or broadband connection. Offloading reduces the total traffic
travelling across the operator’s core network (CN), which means reducing capital investment
and operating expenses for the operator [1]. From operator’s perspective, offloading is done to
relieve congestion in the cellular network and from end user’s perspective, offloading benefits
by making available higher bandwidth and reduction in data service cost.
The remainder of the paper is organized as follows. Sections II, III and IV provide the overview
of femtocells, WiFi and IFW respectively. For each technology, we discuss the offloading meth-
ods, architecture along with challenges and issues in their use and deployment. The concluding
remarks of this work are presented in Section V. The acronyms that are used in the paper are
listed in the Appendix.
II. FEMTOCELL
Femtocell is the technology initially targeted for homes or offices that leverage low power
wireless access points (APs), operating in the licensed spectrum and managed by the cellular
operator. They use the user’s existing broadband connection such as DSL or optical fibre to
connect to the cellular operator’s network. They improve network coverage, capacity and provide
applications not only in homes, enterprises, but also in metropolitan and rural public areas
[2]. Their key features include self-optimization, ease of deployment, IP backhaul and low
power consumption. Femtocells are beneficial in offloading significant traffic from the macrocell
network as it reduces the total traffic travelling over the operator’s wide area radio network.
4. 4
Fig. 2: UTRAN architecture for 3G HNB
Offloading also benefits the non-femtocell subscribers by freeing macro-cellular radio network
resources and providing higher bandwidth for the macro-cellular network users [1].
A. Femtocell Network Architecture
Different cellular systems employ different ways of implementing the actual femtocell archi-
tecture, but there are common requirements regardless of the type of cellular system. A typical
femtocell network is given in Fig. 1. Femto access point (FAP), also Home Node B (HNB) or
HeNB as referred in 3GPP terminology powers the femtocell and is connected to a broadband
router, which is connected to the operator’s CN via cable or DSL and internet providing the
backhaul. Within the femtocell coverage area, the FAP provides internet access to mobile devices
and also provides voice and data services provided by the operator. The Internet Service Provider
(ISP) provides the broadband access which may not be the mobile network operator (MNO).
Femtocell Management System, which is integrated to the MNO’s Network Management System,
manages the femtocells. Fig. 2 illustrates Universal Terrestrial Radio Access Network (UTRAN)
architecture for 3G HNB [6] [7].
HNB: It is Customer Premise Equipment (CPE) that incorporates the capabilities of a standard
Node B and radio resource management functions found within a Radio Network Controller
(RNC). It connects to the UEs via Uu interface and to the HNB Gateway (HNB GW) via Iu-h
interface.
HNB GW: It appears as an RNC to the existing CN using the Iu interface and concentrates
HNB connections. It authenticates and certifies to allow data to and from authorised HNBs.
5. 5
Fig. 3: Local IP Access (LIPA) solution for HNB
HNB GW also provides mechanism to support features like IP based synchronisation and clock
sync distribution.
Iu-h interface: It provides the interface that connects HNB with HNB GW and defines the
security architecture used to provide secure communication over the internet. It supports highly
scalable ad-hoc HNB deployment by defining efficient method for transporting Iu based traffic
and new HNB Application Protocol (HNBAP).
Security Gateway (SeGW): It authenticates and provides HNB with access to HNB Manage-
ment System (HMS) and HNB GW. It establishes Internet Protocol Security (IPsec) tunnels with
HNBs and is responsible for delivering voice, messaging and packet data services between HNB
and CN.
HMS: It facilitates HNB GW discovery and provisions configuration data to the HNB using
TR-069 family of standards. It assigns appropriate serving elements (HMS, HNB GW and SeGW)
after performing location verification of HNB.
Local Gateway (L-GW): It may be co-located with HNB and present when HNB operates
in Local IP Access (LIPA), as described in Section II-B1. L-GW does not use the HNB GW
and has Gn/S5 interface towards the SGSN. In idle mode, it supports the first packet sending
to SGSN and buffers other subsequent downlink packets. The L-GW serves in deactivating the
LIPA connection, which is released with handover.
B. Offloading in Femtocell
Local IP Access (LIPA) and Selected IP Traffic Offload (SIPTO) are the two data offloading
solutions discussed in 3GPP’s System Architecture 2 (SA2) working group [6] [8].
6. 6
Fig. 4: Selected IP Traffic Offload (SIPTO) solution for HNB
1) LIPA: LIPA allows UE to access a residential/enterprise local network that is interconnected
to a HNB. HNB allows home users to access the CN using the local IP backhaul link, which
cause high stress on the CN nodes such as Serving GPRS Support Node (SGSN) and Gateway
GPRS Support Node (GGSN). If however, the IP traffic generated through the HNB is routed
through the ISP’s internet gateway, the stress on the CN nodes could be reduced significantly.
This also provides benefit of accessing other devices like desktops, printers and media servers
within the local subnet without having to go through the CN. In addition, it would also reduce
the number of hops taken by the IP data to reach its destination. 3GPP has enhanced LIPA for
femtocells in Release 10 and is considered a potential method for enabling local IP network
access seamless for home users [9].
Fig. 3 illustrates the Universal Mobile Telecommunications System (UMTS) LIPA architecture
for HNB with S4-UMTS [6]. L-GW can be co-located physically with HNB or as a standalone
entity directly connected to the HNB. In the CN, Serving Gateway (S-GW) and Packet Data
Network Gateway (P-GW) serves for the CN traffic and the sessions for LIPA traffic is managed
by the L-S4 interface between L-GW and SGSN. When the UE is in the idle mode, downlink
packets are buffered in the L-GW and when the UE is in the active mode, the packets buffered
in the L-GW are forwarded on the path between HNB and L-GW [8].
2) SIPTO: SIPTO selectively routes different types of traffic to different destinations originat-
ing from the terminal, which depends on a predefined operator policy. Traffic is offloaded from
the operator’s CN and routed to the internet. So, SIPTO provides access to internet for a UE that
is connected to HNB. It has been developed for two different flavors, namely, Home Network
SIPTO, for traffic originating from HNB and Macro Network SIPTO, for traffic originating from
7. 7
macro node [6].
Fig. 4 describes how SIPTO is achieved in a UMTS/GPRS macro network. The CN includes
the GGSN, which is deployed closer to Radio Access Network (RAN) and above RNC. The
entities in the CN and RAN nodes are not influenced in this architecture. The CN traffic can
utilize the direct tunnel between RNC and GGSN. Operators might perform traffic break-out at
the Gn interface between SGSN and GGSN and redirect RNC’s towards a traffic break-out node
located closer to RAN by using a direct tunnel.
C. Challenges and Issues with Femtocells
1. LIPA introduces several changes to the existing network elements such as SGSN, Home
Subscriber Server (HSS) and RAN. The first change is placing the CN node (P-GW or GGSN)
in the home or enterprise, which adds to the implementation complexity of HNB. Further, to
associate nodes like SGSN, UE and HNB in the customer premises, signalling changes are
required to the existing protocols [9]. Also, SGSN is involved in L-GW selection, which might
prove to be a burden when scaling it for large number of CPE. SIPTO allows traffic break-
out above RAN, which doesn’t reduce congestion on the cell site backhaul or radio interface.
Therefore, “SIPTO for macro” doesn’t help much to solve the problem of capacity crunch [9].
2. Femtocells use the licensed band, for example UMTS band, and it might work on the
same carrier frequency as used by the existing macro infrastructure. As the number of cells in
a network increase, there are more cell borders, leading to increased potential for interference
[10]. Femtocells should be capable of adjusting its transmission to prevent it from interfering in
the macrocell networks as well as its neighbouring femtocell networks.
3. Femtocells should demonstrate self-organizing capabilities to co-ordinate between all cells.
The network should be able to monitor its own performance, source and traffic type and auto-
matically adapting itself to achieve optimal performance. In addition, ease of user installation
and activation is also an important factor. All this demands excellent management of femto radio
resource.
4. Deployment of large number of femtocells would mean the need for large number of wired
and wireless backhaul solutions. Some networks might use wired backhaul whereas some might
use wireless backhaul, which depends on the local constraints [2]. With more backhaul links,
there will be more hubs and congestion points. In order to have the desired performance, careful
8. 8
Fig. 5: 3GPP I-WLAN roaming architecture
planning is required to avoid creating bottlenecks and decreased capacity.
III. WIFI
WiFi is the name for the wireless ethernet 802.11, which is the most popular and widely
deployed standard for Wireless Local Areas Networks (WLANs) [11]. Prevalence of WiFi at
home, various hotspots and also in numerous 3G devices has made WiFi a potential seamless
extension of 3G. WiFi provides benefit of higher data rates to users over the radio interface. WiFi
working on unlicensed spectrum in the 2.4 GHz band, has the potential to reduce interference
to the macro-cellular network. WiFi offload solutions range from the basic data offload straight
to internet based applications to the integrated traffic or data-voice offload using Interworking
WLAN (I-WLAN).
9. 9
A. WiFi Offloading with I-WLAN
I-WLAN is a 3GPP defined standard for interworking between 3GPP networks and WLANs
[12]. It provides network operators with a standardized and controlled approach to offload traffic
over WiFi networks and allowing WiFi traffic to be tethered back to the 3GPP CN. I-WLAN
allows flexibility in deploying secured, automatic and value added WiFi access both in trusted
and untrusted hotspots.
Fig. 5 depicts the 3GPP I-WLAN roaming architecture [12]. The UE attaches to its 3GPP
network for both voice and data services, in the absence of I-WLAN. When UE moves out of its
3GPP coverage or during congestion, UE scans for closest WLAN access network in its vicinity
for data offloading. The UE connects to the WiFi over internet based on its configured policy,
which can be operator or user configured. Operators ensure same user provisioning, authentication
and authorization that already exists for 3G services with ease of integration towards backend
systems [13]. With I-WLAN, authentication is performed transparently to the selected network
without any user intervention. As a form of authentication, UMTS carriers will use Universal
Subscriber Identity Module (USIM) credentials with Extension Authentication Protocol Method
for UMTS Authentication and Key Agreement (EAP-AKA). Request for Access Authentication
via EAP-based messages is sent to the Packet Data Gateway (PDG), after which IPsec tunnel
is established between the UE and PDG [14]. A Mobile IP (MIP) tunnel is set up inside the
IPsec tunnel by MIP agent. Some policies may be included for Access Authentication procedure
for user IP connection to PDG and external IP networks. User’s information is accessed from
Home Location Register (HLR)/HSS and sent to 3GPP Access, Authentication and Authorisation
(AAA) server for the authorization of user’s service subscription. Domain Name System (DNS)
queries for the UE’s remote IP address and the requested WLAN Access Point Name (W-APN) is
resolved by the 3GPP home network. After access authentication, user will have IP connectivity
over the PDG and external IP networks. It should be noted that IP address can be allocated
before or after the authentication process.
After the user switches from 3GPP to WiFi, the 3GPP based services continues to be streamed
over WiFi network without any service interruption. Operator delivers all the 3GPP subscribed
services from the home network to the user and may also push some new services like VoIP,
video sharing within a same billing and charging system [14].
In case the user is roaming, it exits the WiFi network range and upon detection of a 3GPP
10. 10
network, requests for a handover from WiFi to the visited 3GPP network. The UE attaches
to the visited network after a break-before-make handover. The IP address resolution at the
PDG, which is providing access to the selected service will be performed in visited network. A
relation is established between the WLAN Access Gateway (WAG) and PDG, after which WAG
routes data from WLAN access network to visited network. To ensure that the visited network
perform services and charging as provided by the home network, visited network should check
the authorized service subscription information [12]. This is done by the 3GPP AAA Proxy,
which passes subscription information to the home 3GPP AAA server. The decision whether the
visited service is allowed or not is made by the home network based on the user subscription
information, visited network capabilities, roaming agreement, etc.
B. Challenges and Issues with WiFi
1. I-WLAN does not address the aspect of seamless roaming between different type of
networks. It is not possible to initiate a call on a WiFi network and continue the call on
a cellular network. The break-before-make handover between different network types lead to
sudden disruptions, which break the continuity of a call.
2. WiFi offloading works with only those devices that support WiFi. Even if WiFi is going
popular these days, not all devices available in the market are WiFi enabled. There are still some
markets which do not provide WiFi enabled handsets.
3. WiFi works in the unlicensed band due to which there is lack of scalability and quality
of service (QoS). It cannot control potential interference from other WiFi service providers or
other RF resources that are sharing the same RF spectrum. Some other devices that use the
2.4 GHz band are microwave oven, Zigbee and Bluetooth. Therefore, WiFi face a challenge
for scaling the service with increasing competition between multiple and overlapping multiple
service providers. Efficient resource allocation of unlicensed spectrum should be facilitated with
possible protocols to enable the management of QoS more effectively.
4. WiFi networks have limited range. A typical WiFi access point would have a range of 32 m
indoors and 95 m outdoors. Due to the reach requirement of WLAN devices, it experiences fairly
high power consumption compared to other standards [11]. Due to the high power consumption,
using WiFi on handsets drains away a lot of battery power.
11. 11
Fig. 6: Residential provisioning systems of femto and WiFi networks
IV. INTEGRATED FEMTO-WIFI
Both WiFi and femtocells are the technologies meant for small area coverage. WiFi provides
only internet access, whereas femtocell networks provide access to the internet as well as
other access services provided by the network operator. At first, it appeared that these two
technologies are competing technologies and cannot co-exist. However, recent developments
have come forward showing that these two technologies actually complement each other and
benefits can be reaped by aggregation of both 3G and WiFi [5]. The strategy of integrated
femto-WiFi is to get the best out of both worlds.
With only femto or WiFi, gross offload or bulk offload is experienced, where all the data
is handled by only cellular or WiFi alone. Whereas IFW will lead to smart and fine-grained
offload. ‘Smart’ because offloading criteria can be based on user, application type, traffic type,
protocol type, network condition, etc., not just based on coverage. For example, offloading policy
may depend of the type of user subscription or whether the traffic is VoIP, video or whether
the protocol is FTP, TCP, etc. ‘Fine-grained offload’ because handover of single flow from one
radio to another can be done keeping all other flows on the first radio. This provides granular
offload which will help in achieving load balancing as well. The idea of IFW is beneficial in
terms of energy savings as well [15]. Keeping WiFi ON all the time on the user handsets drains
away significant battery power. So with IFW, we can achieve femto awareness, i.e., when user
enters the coverage area of femto, and if it is femto and WiFi together, then FAP can trigger to
turn ON the WiFi radio on the UE without user intervention.
Initially, the focus of femto and WiFi were for indoor residential deployment, but soon it was
considered for enterprise and metro deployment as well. Therefore, the IFW networks can be
considered for deployment scenarios of residential, enterprise, metro and public rural areas. We
12. 12
discuss only the residential provisioning system for the IFW network based on UMTS standard
as shown in Fig. 6 [5].
The IFW AP is connected to a Residential Gateway such as broadband modem and router,
which in turn is connected to Post Office Protocol (POP) in the ISP network and SeGW in the
operator’s network. SeGW is further connected to the Femto Gateway (FGW) and operator’s
CN. The femto data flows from the IFW AP to the SeGW via IPSec tunnel over the Iu-h and
further towards the FGW. FGW interacts with the operator’s CN through the Iu-CS interface for
circuit switched data and Iu-PS interface for packet switched data. Residential GW Provisioning
Servers are different subsystems in the operator and ISP networks. TR-069 over SSL/TLS or
IPSec is used to provision the femto part and TR-069 protocol is used to provision the WiFi
part of the IFW AP.
IFW networks include the femto, WiFi and the broadband backhaul part, and the various
players who manage these parts are network operator, ISP and end user. Different operational
scenarios is possible, based on which part is managed by whom. The most common operational
scenario is when the network operator offers the IFW device, broadband operated by ISP and
WiFi provisioning and management is left to the user’s desire. This scenario allows flexibility
of configuring WiFi and home LAN to the user and also reduces cost for WiFi user support.
A. Traffic Flow Management
In this section, we discuss different traffic flow management operations possible with the
integration of femto and WiFi. First, we discuss the control plane aspects, then we discuss
offloading from femto to WiFi radio interface, which involves communication in idle mode. We
describe femto-WiFi handover, which refer to moving active sessions from one radio to another
with session continuity. Then, we describe simultaneous flow management where individual IP
flow may be moved from femto to WiFi or vice-versa keeping the remaining IP flows intact on
the first radio.
1) Control Plane Aspects: WiFi networks are unplanned networks and if operators want to
include WiFi as part of their service, it is important how the operators control the management
of radio resources. An algorithm is needed to provide operator’s policy to UE for unplanned
networks. Also, the UE should have a mechanism to detect WiFi networks and use the WiFi
service according to the operator’s policy. This is achieved by 3GPP standardized Access Network
13. 13
Discovery and Selection Function (ANDSF) [16].
ANDSF is a layer 3 protocol specified by 3GPP and assists UE to discover non-3GPP access
networks such as WiFi. In the presence of two or more different types of networks, ANDSF
enables UE to select the most appropriate access network by exchanging required information
between the UE and server [16]. It enables MNO to define and modify policies for WiFi
discovery, Wi-Fi offload and simultaneous femto/cellular and WiFi management. These policies
are contained in ANDSF Management Object (MO) which are stored in UE and ANDSF server.
Open Mobile Alliance Device Management (OMA DM) protocols can modify and synchronize
the MOs.
Discovery Information: A list of networks available and information around the vicinity of UE
is made available to expedite the connection to these networks. In case of WiFi, the information
includes WiFi service set identifier (SSID), security modes (e.g. WPA2), mode of operation,
hidden APs.
Inter-System Mobility Policy (ISMP): It defines policies for WiFi offload and allows MNOs
to set rules on the UE for a decision on which access technology and network to give priority.
The validity of the policy should take into account validity area, roaming, geolocation of UE,
CELL ID of cellular network, WLAN SSID and time of the day. A policy is valid if all the
conditions match. After an available network is identified by the ANDSF policy then the highest
priority rule becomes ‘active’ and network is selected [17].
Inter-System Routing Policy (ISRP): It defines policies for simultaneous cellular and WiFi
management and allows operator to provide policies based on the UE’s traffic exchange. IP
Flow Mobility (IFOM), a seamless offload rule provided by ISRP, allows on a per flow basis to
select the access on which each IP flow should be supported and seamlessly move them between
accesses [5]. The packets may be routed based on validity area, time of day and Access Point
Name (APN) of the network. Source and destination IP address and ports classify the traffic
flow.
A client server based protocol called Dual Stack Mobile IP (DSMIP) was introduced in 3GPP
Release 8 [18]. DSMIP enables seamless offload between 3G and WiFi and provides IP address
preservation for IPv4 and IPv6 sessions, which means that when IP flows from one access to
another, the process is transparent to the user leading to uninterrupted user experience.
DSMIP allows seamless offload solution where all the traffic is routed to WLAN. In cases
14. 14
Fig. 7: WiFi offloading architecture
where we want to offload only selective IP flow to WLAN and other traffic still maintained over
the access network, we need to allow simultaneous registration of multiple addresses, which
could be achieved by DSMIPv6 extensions. The protocol extension is known as IFOM [19] and
it allows operators to specify how the IP flows are routed through the access network and to
offload some selective traffic (e.g. best effort traffic) to WLAN while still using access network
for other traffic (e.g. traffic requiring specific QoS).
2) Smart WiFi Offloading: We refer here WiFi offloading as directing traffic flow from femto
to WiFi radio interface with minimal user intervention. Some important elements to consider
during WiFi offloading are discovery of WiFi AP’s, selection of appropriate WiFi network and
connection to the WiFi network [5]. Firstly, the UE needs to discover the WiFi APs around
its vicinity and appropriate AP has to be selected which could be based on network capability
such as QoS, signal strength connectivity, etc., or based on a list of preferred WiFi access
networks. The user or the MNO might be able to configure the list of preferred WiFi access
networks for automatic selection. ANDSF procedure is used in case the MNO is responsible for
the provisioning of discovery and selection of the preferred WiFi network [16].
Fig. 7 illustrates the WiFi offloading architecture using ANDSF. The ANDSF Context Infor-
mation Collection System is responsible for gathering the context information of AP and RNC
and sending it to the ANDSF server [17]. The policy manager updates and manages the MNO’s
policy in ANDSF server.
3) Seamless Femto-WiFi Handovers: The term handover refers to the process of moving an
active session from femto to WiFi radio and vice-versa. During handover, the session continuity
15. 15
is very important, therefore the IP address of the UE must be preserved as the UE links from
one radio to another.
At IP level, there are two types of handover protocols, namely, Network-Based-Mobility
(NBM) protocols and Host-Based-Mobility (HBM) protocols [5]. In NBM protocols, only net-
work is involved in mobility management, with only minimal involvement from UE. PMIP,
PMIPv6 and GTP are examples of NBM protocols. HBM protocols, on the other hand, both UE
and network participate in the mobility management. MIP, MIPv6 and DSMIv6 are examples of
HBM protocols [20]. DSMIP allows make-before-break mobility across networks, which helps
maintain the IP address continuity and seamless handover between cellular network and WLAN
[5].
4) Simultaneous Femto-WiFi Flow Management: Section IV-A2 and IV-A3 described the
flow management when UE at any time, is connected to a single radio, either femto or WiFi.
Additional functionalities like flow segregation, flow mobility and flow aggregation is possible
when UE support simultaneous connection with both femto and WiFi [5]. The functionalities
can be implemented using HBM or NBM protocols with certain extensions given in [18].
In flow segregation, the most appropriate radio interface is selected for sending different type
of application flows. The choice of radio interface is decided in terms of cost, bandwidth, etc.
For instance, VoIP may be directed over the femto or cellular network, whereas http traffic or
third party application might be sent over WiFi link.
Flow mobility refers to moving individual application flow from one radio link to another,
while retaining the rest of the flow on its original link. For example, if WiFi link is experiencing
reduced capacity due to increasing interference then the http traffic flow might be moved from
WiFi to femto/cellular link, while the third party applications continue to flow on WiFi link.
In flow aggregation, a single application flow is split into two different sub flows and sent
simultaneously across both femto and WiFi. This helps in dynamic load balancing between the
two radio links and also provides an alternate route in case of a radio link failure as communi-
cation becomes possible across the surviving link. It also benefits demanding applications such
as HD video, by increasing the overall bandwidth.
16. 16
Fig. 8: Core network based IFW function
Fig. 9: Edge based IFW function
Fig. 10: Gateway based IFW function
17. 17
B. Architectural Aspects
The IFW network architecture depends on the specific deployment and operational scenario.
Nevertheless, some key aspects are common to all scenarios. Both UE and network will have an
impact by the integration of femto and WiFi. Operator policies will need some kind of Policy
Client such as ANDSF client, whereas, UE should support some kind of Connection Manager.
In case of impact on network, multiple choices may exist as IFW functionality could reside in
one of the network nodes.
As the question arising here, where is the integration going to take place, we could have
three options as described next. First option is, IFW functionality residing at the GGSN of CN.
Second, the integration point could be the AP itself or third, in between CN and AP, i.e., at the
gateways. These are illustrated in Figs. 8, 9 and 10, based on UMTS CN [15].
1) GGSN Based IFW Functionality: The UE is connected to the FAP and WiFi AP, which are
collocated in an integrated environment (possibly in a same box/physical entity). The integrated
AP is connected to the Broadband Modem (BBM) which is further connected to the CN through
the femto path and WiFi path.
The femto path comprises of FGW, SGSN and GGSN. The internet bound traffic may be
offloaded from the Traffic Offload Function (TOF) along Macro-SIPTO path or directly from
the Broadband IP Network (BBIPN) along the Femto-SIPTO path. The WiFi path comprises
of Tunnel Terminating Gateway (TTG), which establishes secure IPsec tunnel between UE and
TTG over BBIPN. The traffic towards the CN from TTG can be diverted to the Operator Service
Network through GGSN. The internet traffic can be offloaded to internet through TOF. The LIPA
configuration is also shown in the figure, where the local traffic is diverted to the local IP Network
by the Local Gateway (LGW).
As the femto path and WiFi path converge at GGSN, IFW function (IFW-F) would reside
at the GGSN network node. In case of DSMIP based handover and flow management, IFW-F
includes Home Agent Functionality and while considering PMIP based implementation, IFW-F
includes Local Mobility Agent (LMA) functionality.
The core based IFW-F doesn’t exploit the fact of a presence of femto link, and is a feasible
choice for only macro cellular and WiFi integration, when the cellular base station and WiFi AP
are not collocated and separated geographically. Here, existing solution for IP flow management
such as I-WLAN can be used.
18. 18
2) AP Based IFW Functionality: As the femto and WiFi AP are collocated in IFW network,
the IFW functionality can be located in nor near the IFW AP. This is edge based IFW function-
ality and as the signalling doesn’t need to traverse to the CN, this architecture is attractive to
the network operator.
3) FGW Based IFW Functionality: The third variation is residing the IFW functionality
at the FGW, which is also integrated with WiFi gateways, i.e., TTG and PDG. Gateway has
wide visibility of multiple APs and can therefore perform clever management for network level
optimization [15].
C. Challenges and Issues with IFW
1. Integrating femto and WiFi would require proper joint service provisioning and resource
management methods between the two technologies. In case of residential and enterprise de-
ployment scenario, the two technologies are managed separately, so provisioning in this case,
could be performed separately. However, in case of metro deployment, it is beneficial to have
a single provisioning process, as the operator manages both femto and WiFi together [15].
A single network management is also beneficial in such case for configuration management,
fault management and performance management of both technologies. Also, in order to reduce
the amount of manual operations in the deployment of new provisioned AP device, device
management process is preferred to be automated.
2. Users may have different billing agreement with different access networks and it might not
be clear who owns the user experience. This demands a consistent billing system between femto
and WiFi. Also, the network operator may have little control over ISP’s policy for prioritizing
QoS dependent services [5]. Therefore, an effective joint policy control is required for good user
experience in the IFW environment.
3. Integration of femto and WiFi will have to deal with various security issues. For instance,
customers in enterprise deployment scenario might have particular security requirements that
might be difficult for the network operator to fulfill. In fact, it will be a challenging task for the
network operator to work together with the enterprise firewalls.
4. IFW work in close interaction with different layers of the hierarchical HetNets, which makes
it important to develop open standard interfaces. This would allow different vendor products to
be used in different parts of the network so that different tasks could benefit from the best
19. 19
available product.
5. In order to have efficient and intelligent implementation, it is important to identify infor-
mation exchange needed between different individual APs [5]. It becomes essential to define
Application Programming Interfaces (APIs) for such information exchange to achieve interop-
erability between APs.
V. CONCLUSION
With the extreme success of mobile broadband and takeup of smartphones, tablets and other
data devices, there is increasing demand for data traffic to be carried on mobile networks today,
leading to high congestion on the macro cellular networks. Traffic offloading plays an important
role in easing the pressure from the macro-cellular core networks leading to increased capacity
and better user experience.
This paper provided a comprehensive overview on various traffic offloading solutions. Fem-
tocell technology is a compelling solution, which provides the benefits of scalability, automatic
configuration and self optimization. Another alternative is WiFi, which is widely available at
homes and hotspots. Further, we discussed integrating femtocells and WiFi together in order
to reap the benefits of both femtocell and WiFi technology. It is recommended that the future
investigation and research on integrated femto-WiFi should focus on joint service provisioning
and resource management, joint policy matters, interoperability implementation aspects, security
issues, etc., which would possibly solve the current high traffic demand to some extent and
provide better user experience.
REFERENCES
[1] Femto Forum,“Femtocells - Natural Solution for Offload,” White Paper, Jun. 2010.
[2] Small Cell Forum Ltd., “Small cells - what’s the big idea? Femtocells are expanding beyond the homes,” White Paper,
Feb. 2012.
[3] Qualcomm Inc., “Traffic Management Strategies for Operators,” White Paper, Jan. 2011.
[4] Femto Forum, “Wireless in the home and office: The need for both 3G femtocells and WiFi access points,” White Paper,
Jan. 2010.
[5] Small Cell Forum Ltd., “Integrated Femto-WiFi (IFW) Networks,” White Paper, Feb. 2012.
[6] “3GPP TR 23.829 version 10.0.1 Release 10 Technical Specification Group Services and System Aspects; Local IP Access
and Selected IP Traffic Offload,” Tech. Rep., Oct. 2011.
[7] J. Chen, P. Rauber, D. Singh, C. Sundarraman, P. Tinnakornsrisuphap, and M. Yavuz, “Femtocell Architecture and Network
Aspects,” Qualcomm, Jan. 2010.
20. 20
[8] K. Samdanis, T. Taleb, and S. Schmid, “Traffic Offload Enhancements for eUTRAN,” IEEE Commun. Surveys & Tutorials,
vol. PP, no. 99, pp. 1–13, Sep. 2011.
[9] WirelessE2E LLC, “Analysis of Traffic Offload: WiFi to the Rescue,” White Paper, Sep. 2010.
[10] V. Chandrasekhar, J. Andrews, and A. Gatherer, “Femtocell Networks: A Survey,” IEEE Commun. Mag., vol. 46, no. 9,
pp. 59–67, Sep. 2008.
[11] Intel Corporation, “Wireless Networking: Deployment Considerations for Today and Tomorrow,” White Paper, 2003.
[12] “3GPP TS 23.234 version 10.0.0 Release 10, Universal Mobile Telecommunications System (UMTS); LTE; 3GPP system
to Wireless Local Area Network (WLAN) interworking; System description,” Tech. Rep., Oct. 2011.
[13] H. Wu, Z. Zhong, and H. Huang, “A fast authentic handover scheme for WLAN-3GPP interworking network,” in Proc.
ICSPCS, Sep. 2009, pp. 1–6.
[14] Managing data offloading securely over WLAN access networks with I-WLAN. Available online at:
http://www.greenpacket.com, retrieved 10 Apr. 2012.
[15] Presentations. Available online at: http://www.smallcellforum.org/resources-presentations, retrieved 10 Apr. 2012.
[16] “3GPP TS 24.312 version 10.5.0 Release 10, Universal Mobile Telecommunications System (UMTS); LTE; Access Network
Discovery and Selection Function (ANDSF) Management Object (MO),” Tech. Rep., Mar. 2012.
[17] GSM Association, “WiFi Offload,” White Paper, Apr. 2010.
[18] Qualcomm Inc., “3G/WiFi Seamless Offload,” White Paper, Mar. 2010.
[19] “3GPP TS 23.261 version 10.2.0 Release 10, Universal Mobile Telecommunications System (UMTS); LTE; IP flow mobility
and seamless Wireless Local Area Network (WLAN) offload; Stage 2,” Tech. Rep., Mar. 2012.
[20] A. Udugama, M. Iqbal, U. Toseef, C. Goerg, C. Fan, and M. Schlaeger, “Evaluation of a Network Based Mobility
Management Protocol: PMIPv6,” in Proc. IEEE VTC, Apr. 2009, pp. 1–5.
21. 21
APPENDIX
TABLE I: List of Acronyms
3G Third Generation GGSN GPRS Support Node
3GPP Third Generation Partnership
Project
GPRS General Packet Radio Service
4G Fourth Generation GTP GPRS Tunneling Protocol
AAA Access, Authentication and Autho-
risation
HBM Host Based Mobility
ANDSF Access Network Discovery and Se-
lection Function
HetNet Heterogeneous Network
AP Access Point HLR Home Location Register
API Application Programming Interface HMS HNB Management System
APN Access Point Name HNB Home Node B
BBIPN Broadband IP Network HNB GW HNB Gateway
BBM Broad Band Modem HNBAP HNB Application Protocol
CN Core Network HSS Home Subcriber Server
CPE Customer Premise Equipment I-WLAN Interworking WLAN
DNS Domain Name System IEEE Institute of Electrical and Electronic
Engineers
DSL Digital Subscriber Line IFOM IP Flow Mobility
DSMIP Dual Stack Mobile IP IFW Integrated Femto Wifi
EAP-AKA Extension Authentication Protocol
Method for UMTS Authentication
and Key Agreement
IP Internet Protocol
FAP Femto Access Point IPSec Internet Protocol Security
FGW Femto Gateway ISMP Inter-System Mobility Policy
FTP File Transfer Protocol ISP Internet Service Provider
22. 22
ISRP Inter-System Routing Policy SeGW Security Gateway
L-GW Local Gateway SGSN Serving GPRS Support Node
LAN Local Area Network SIPTO Selected IP Traffic Offload
LIPA Local IP Access SLF Serving Location Function
LMA Local Mobility Agent SSID Service Set Identifier
MIP Mobile IP SSL Secure Sockets Layer
MNO Mobile Network Operator TCP Transmission Control Protocol
MO Management Object TLS Transport Layer Security
NMB Network Based Mobility TOF Traffic Offload Function
OCS Online Charging System TTG Tunnel Terminating Gateway
OMA DM Open Mobile Alliance Device Man-
agement
UE User Equipment
P-GW Packet Data Network Gateway UMTS Universal Mobile Telecommunica-
tions System
PDG Packet Data Gateway USIM Universal Subscriber Identity Mod-
ule
PMIP Proxy Mobile IP UTRAN Universal Terrestrial Radio Access
Network
POP Post Office Protocol VoIP Voice over IP
QoS Quality of Service W-APN WLAN Access Point Name
RAN Radio Access Network WAG WLAN Access Gateway
RF Radio Frequency WiFi Wireless Fidelity
RNC Radio Network Controller WLAN Wireless Local Area Networks
S-GW Serving Gateway WPA Wi-Fi Protected Access
SA2 System Architecture 2