Cellular Wi-Fi IntegrationA comprehensive

analysisPart I
Prabhakar Chitrapu, Alex Reznik, Juan Carlos Zuniga - July 16, 2012

From Competitive Positions to Shared Goals

Cellular and Wi-Fi radio technologies originated and evolved from two fundamentally different
objectives. The former was motivated by the desire to make telephony technology mobile, and the
latter by the desire to make data communications wireless. Over time, each has trended towards the
other, with wireless data a central use of cellular technology today while over-the-top services
provide voice over data networks. This confluence seems headed towards a fundamentally
integrated cellular and Wi-Fi landscape, but the evolutionary nature of the trend has resulted in a
broad variety of approaches and solutions.
Significant advancements in data over wireless began to emerge in the early 1980s. InterDigital
helped pioneer this capability, with TDMA access in a DSP-based implementation in the 1980s laying
the technological blocks for the digital wireless revolution of the 1990s, followed by InterDigitals
WCDMA system using CDMA in channels as wide as 20 MHz to provide high-capacity fixed wireless
access. With the introduction of the ETSIs GSM system (a digital TDMA system with DSP-based
devices), the vision of ubiquitous connectivity was at last beginning to appear real. By the late
1990s, ETSI and then 3GPP were already looking towards providing packet connectivity, first as the
GPRS add-on system to GSM and then, with UMTS (a CDMA system), as a fundamental capability.
At the same time a completely different vision of wireless access emerged. This vision was based on
the success of Local Area Networks particularly Ethernet in connecting the enterprise intranet
into a highly capable local network. Using the free-for-all ISM bands, first allocated by the US
FCC in 1985, the IEEE 802.11 Working Group began looking at taking the Ethernet wireless. With
no reliance on expensive licensed spectrum and fewer concerns regarding QoS, Wi-Fi technology
emerged as a strong consumer and, increasingly, enterprise wireless solution by the mid-2000s.
The initial response from the cellular community was purely defensive. Rather than embracing and
integrating the new technology, 3GPP embarked on a fundamental re-architecture of its system
which resulted in a completely IP-based packet-focused Evolved Packet Core (EPC). Combined with
OFDM-based LTE access technology, this was supposed to be the answer to the threat posed by WiFi.
The response from the Wi-Fi community to the customer demand for integration with mobile
communication solutions was similarly lukewarm. With the MIP family of IP enhancement, IETF
took the lead in trying to provide mobility support to IP-based systems such as Wi-Fi. However, the
limited traction these protocols did find was in 3GPP where MIP and PMIP have been adopted as
solutions for the EPC. Until very recently the focus of the Wi-Fi community remained on delivering
access to ever-greater bandwidth (with 802.11n), while the need for QoS management and mobility

was completely ignored.

That mutual defensiveness has now reversed, fuelled both by consumer demand regarding data to
new device types and operator efforts to relieve network pressure. From the cellular side, small,
localized cells, such as Wi-Fi AP coverage regions, are growing in importance. As an IP-based
system, the EPC is well positioned for integration of multiple heterogeneous access technologies and
3GPP is now moving towards taking advantage of EPC to delivery solution for real integration of
technologies such as Wi-Fi. With ANDSF, an integrated policy-based management system for how
devices access spectrum 3GPP has already taken a first step in that direction.
On the other hand, the Wi-Fi community has finally acknowledged that when using mobile devices
(for example, the iPhone), consumers expect to receive the same quality of service whether they use
Wi-Fi, 3G or LTE. By extension, Wi-Fi must provide operators with the tools to manage Wi-Fi
networks in the same way that they can manage their own 3G or LTE networks, and the recent
Hotspot 2.0 profile and the associated PassPoint certification program sees the Wi-Fi Alliance
beginning to move towards delivering such carrier-grade Wi-Fi solutions to the market.

Finally, the trend towards true integration is beginning to come to the fore. The need is particularly
acute in small cells designed to address the high spectrum needs of local consumer and enterprise
networks and Hotspots. Thus, through its work on Integrated Femto-Wi-Fi (IFW), the Small Cells
Forum is already taking steps towards defining the near-future of such spectrum integration. What
will this future hold? A combination of integrated small-cell solutions, smart connection management
at the terminal, within a policy framework that provides management controls to both users and
operators.
Wi-Fi Evolution
Initially conceived in the late 1980s as a wireless extension of Ethernet, initial WLAN installations
used the then recently FCC established unlicensed frequency bands and were primarily confined to

fixed enterprise deployments. With the establishment of the 802.11 Working Group by the Institute
of Electrical and Electronics Engineers (IEEE) in 1991, the increasing speeds of 802.11a and
802.11b (operating on the unlicensed 5 and 2.4 GHz bands respectively), and the Wireless Ethernet
Compatibility Alliance (WECA) coining the name Wi-Fi, and initiated an industry marketing,
interoperability and certification program in 1999, Wi-Fi was successfully launched as a broadlyadopted wireless standard.
Close coordination between the IEEE 802.11 Working Group and the Wi-Fi Alliance has continued to
improve capabilities, and newer Wi-Fi versions like IEEE 802.11g (a 54 Mbps version of 802.11b)
and IEEE 802.11n (~600 Mbps at 5GHz using a wider bandwidth and multiple antennas for
transmissions and reception), coupled with backwards compatibility with the older versions of
802.11, has resulted in Wi-Fi products certified as 802.11a/b/g/n. It is important to recognize that
an IEEE 802.11 device is not Wi-Fi unless and until it has been certified by the WFA.
Because Wi-Fi was originally conceived as a simple extension of an Ethernet cable, its design
provided for short-range, local area coverage, and did NOT address general network considerations
such as radio measurements and statistics, device management, or Quality of Service (QoS)
(although IEEE 802.11 did produce standards such as 802.11e for QoS, 802.11i for issues like
improved security). With the relatively recent incorporation of Wi-Fi into most cell phones and
adoption by many mobile operators, some of these considerations are starting to be addressed as
will be examined later in this paper in the Present Activities section.
Mobility Solution Toolbox from IETF
IETF (Internet Engineering Task Force) is the technical body that defines standards according to
which the internet operates. Its standards are the so-called RFCs (Request for Comments), which
cover various aspects of Inter-Network Transport and higher layer functionalities, in terms of a
variety of protocols. For example, IETF developed the IP (Internet Protocol), which defines the
structure of information packets and how they are transported between two end-to-end IP-devices
across an interconnection of networks. IP protocols are designed to be agnostic to the underlying
characteristics of any of the intervening networks and have provided the highway architecture for
the modern internet. It is precisely this independence from the underlying network characteristics
that made IP a natural choice for interworking Cellular and Wi-Fi Networks, since It is a common
language that be supported by both networks.
The basic IP-protocols did not address mobility of the end-devices and these are handled in a series
of Mobile-IP standards. These can be grouped in two main categories: client-based and networkbased solutions. Client-based solutions require some special functionalities in the client device, and
make use of a mobility agent in the network, whereas network-based solutions rely on the network
for both agent and client functionalities, thus making the mobile device agnostic to these mobility
functions and therefore simpler to implement. One of the main goals of any of these mobility
protocols is to provide seamless mobility as the device moves from network to network. This is
essentially achieved by preserving the IP address of the mobile device via the concept of Home IP
Address (which stays invariant) and associated Care-Of IP Address (which changes due to mobility).
The main client-based approach used to provide seamless mobility is based on the Mobile IP (MIP)
protocol [RFC 6275], which lately has been extended into the Dual-Stack Mobile IP (DSMIPv6)
architecture [RFC 5555]. The main network-based approaches are based on the Proxy Mobile IP
(PMIP) protocol [RFC 5213], also an extension of the MIP protocol.
Cellular Wi-Fi
MIP: Mobile IP Protocol

MIP supports uninterrupted routing of IP-packets to and from a mobile device and provides session
continuity by means of a Home Agent (HA), which is an entity located at the Home Network of the
mobile device (also referred to as a Mobile Node - MN) that anchors the permanent IP address
assigned to the mobile device, known as Home Address (HoA). The HA keeps the devices HoA when
the MN has moved from the home network, and redirects traffic to the devices current location. The
HA is informed of the current location by the MN, using a temporary IP address or Care-of Address
(CoA) that the MN acquires from the visited network. A bi-directional IP-tunnel between the MN and
the HA is then used to redirect traffic between these nodes.
MIP, defined originally for IPv4 devices and networks, was subsequently extended to MIPv6 to be
applicable to IPv6 devices and networks. DSMIPv6 is a further extension of MIPv6, where the basic
mobility functionality is extended to also support dual stack IPv4/IPv6 devices and networks.
Accordingly, DSMIPv6 extensions are defined to also register IPv4 addresses and transport of both
IPv4 and IPv6 packets over the tunnel between the HoA and the visited network. These extensions
enable the mobile device to roam between IPv4 and IPv6 access networks seamlessly, and are
considered crucial as IPv4 networks and devices gradually evolve to IPv6.
PMIP: Proxy MIP
PMIP and its IPv6 extension, PMIPv6, are examples of Network-based IP mobility solutions, which
manage the mobility of the mobile device entirely to the network. In this way, the device is not
required to perform any signaling or updates, as it changes of its point-of-attachment (i.e. visited
network) due to its mobility. Hence, these changes become transparent to the mobile terminals IP
protocol stack, resulting in simpler device solutions than those based on baseline MIP.
The PMIP-enabled mobile IP network architecture consists of a central entity, called Local Mobility
anchor (LMA), and a number of Mobile Access Gateways (MAGs), which together define a mobility
domain. The LMA plays the role of a local HA (as in DSMIP networks) and anchors the IP prefixes
used by the MNs. MAGs reside close to the mobile node, usually in the Access Routers (which in turn
are either collocated with the Access Points or directly connected to them).
Detection of movement of mobile devices as well as implementing associated signaling is done by
the MAGs. Typically, the MAG detects mobility through standard terminal operations, such as router
and neighbor discovery or by means of link-layer support, without any mobility specific support from
the device. Bi-directional tunnels between the LMA and the MAGs are set up, so that the mobile
device is able to keep the originally assigned IP address within the mobility domain despite any
location changes. Since the LMA is aware of the actual location of the mobile device, any packets
addressed to the device are tunneled to the appropriate MAG, relieving the mobile device of the
need to manage the IP packet routing due to its own mobility.
Extensions to Mobility of Multi-connection Devices
The underlying assumption in basic MIP and all their derivatives (such as DSMIP [RFC5555], an
extension of DSMIP to support simultaneous IPv4/IPv6 operation) is that a mobile device has a single
Home Address and a single Care of Address (which may change due to MN mobility). However,
modern devices, such as smart phones, can support multiple IP connections, for example via Cellular
and Wi-Fi network interfaces. Clearly, the DSMIP cannot support mobility of such devices, and IETF
standardized the basic components required to remove such limitations. These components are:
multiple care-of address registration support [RFC 5648], flow bindings support [RFC 6088], and
traffic selectors definition [RFC 6089].
Multiple care-of-address registration enables a device with multiple IP connections to be registered

with a single Home-Address and multiple Care-of-Addresses. This allows the management of
mobility of one or more network connections, by updating the corresponding care-of-addresses with
the Home Agent. Flow bindings concepts enable the association (or binding) of individual IP-Flows to
specific care-of-addresses (or network interfaces). IP-flows in turn are defined by the notions of
traffic selectors, defined in RFC6089. These concepts extend the concept of mobility to individual
IP-Flows and allow one to move IP-Flows dynamically from one network interface to another (e.g.
mobility of IP-Flows from Cellular to Wi-Fi and vice versa).
In the case of network-based mobility solutions, the mobility management control is not located in
the mobile device but in the network. PMIPv6 and GTP are network-based mobility protocols and
they provide some basic support for handling multiple interfaces. However, they do not support
mobility at an IP flow granularity. For this reason, extensions are being defined in IETF [ IETF draftietf-netext-pmipv6-flowmob, IETF draft-ietf-netext-logical-interface-support ] and the recently
approved 3GPP study item on network-based flow mobility (NB-IFOM). The first extension allows a
mobile device being attached to two different media access gateways (MAG) with two different
interfaces (e.g. cellular and Wi-Fi) in the same PMIPv6 domain. The second extension allows a MAG
to forward traffic to a mobile device, even if the IP address (e.g. IPv6 prefix) was originally delegated
to the mobile via a different MAG.
Extensible Authentication Protocols (EAP)
Extensible Authentication Protocol (EAP) is a simple authentication protocol defined in [RFC 3748]
which is designed to provide a generic framework for user authentication in a network within IPNetworks. A key aspect of EAP is that it itself does not define how authentication is done rather it
defines a framework which can be used to define a specific authentication protocol. Such protocols
are referred to as EAP methods and in most cases the protocol involves authentication with a remote
server.
Examples of such EAP methods include EAP-SIM [RFC 4186] and EAP-AKA [RFC 4187, RFC 5448],
which are used for authenticating cellular subscribers over Wi-Fi networks. Specifically, these use
the SIM or USIM credentials of cellular subscribers for authentication. These protocols also provide
for mutual authentication, meaning that UE is authenticated to the mobile network, but, at the same
time, the UE is able to verify the identity of the network as well.
As an example, a Wi-Fi network performing EAP-SIM authentication would consist of a Wi-Fi Access
Point (AP) to which the UE is attached. The AP is also connected to an AAA-server, which in turn is
connected to a mobile operators HLR/HSS for the purpose of authenticating the user. The AP asks
the UE to identify itself via EAP message exchange. On hearing back from the UE, the AP passes on
the UE responses to the AAA Server via RADIUS messages, which are further passed on to
HLR/HSS. Upon successful authentication, the HLR/HSS informs the AAA Server, which in turn
communicates the results to the AP and eventually to the UE.
This concludes a review of efforts by the internet community to integrate cellular technology. In the
second part of this paper, well look at the reverse: how the GSM community and 3GPP have worked
to integrate Wi-Fi.
About the Authors
Prabhakar Chitrapu is a Senior Principal Engineer at InterDigital Communications in King of
Prussia, PA. He has contributed to various aspects of 3G/4G cellular and WLAN networks and been
an active participant in industry forums, such as GSMA, UMTS Forum and Femto Forum (including
Vice-Chair of the WLAN-Interworking Task Force at GSMA, Vice-Chair of the Working Group on

Networks, and Champion of the Integrated Femto-WiFi Networks work item at the Femto Forum).
Prior to his 20-year corporate career, Prabhakar was Assistant Professor at Drexel University. He
holds a PhD degree from Delft University, The Netherlands.
Alex Reznik is a Senior Principal Engineer at InterDigital, currently leading the companys research
and system design activities in the area of IP mobility and heterogeneous networks. Since joining
InterDigital in 1999, he has been involved in a wide range of projects, including leadership of 3G
modem ASIC architecture, design of advanced wireless security systems, and coordination of
standards strategy in the cognitive networks space. He earned his B.S.E.E. Summa Cum Laude from
The Cooper Union, S.M. in Electrical Engineering and Computer Science from the Massachusetts
Institute of Technology, and Ph.D. in Electrical Engineering from Princeton University.
Juan Carlos Zuniga is a Member of the Technical Staff at InterDigital, in Montral Canada, where
has worked in the standardization and development of advanced wireless access technologies since
2001. He previously worked at Harris Communications in Canada, Nortel Networks in the United
Kingdom, and Kb/Tel in Mexico. Throughout his career he has participated and held leadership
positions in several standardisation bodies including IETF, 3GPP, IEEE 802.11, 802.16, and 802.21.
He received his engineering degree from the UNAM, Mexico, and his M.Sc. and DIC from the
Imperial College London, UK.