intelligent connection manager for seamless interworking of multi-technology mobile devices
TRANSCRIPT
◆ Intelligent Connection Manager for SeamlessInterworking of Multi-Technology Mobile DevicesEdward Grinshpun, David W. Faucher, and Sameer Sharma
As wireless service providers ramp up their Long Term Evolution (LTE)installations, and telecommunications and cable operators continue toexpand Wi-Fi coverage, 100 percent wireless coverage with overlappingareas of 3G, 4G, and Wi-Fi access is becoming a reality. Ensuring that a multi-mode device can roam across different radio access technologies whilemaintaining seamless session and service continuity across virtual privatenetwork (VPN), Voice over Internet Protocol (VoIP), and streaming videoapplications is critical. Differences in 3GPP, 3GPP2, and WiMAX standards ledto the absence of uniform interworking solutions. Specific interworkingstandards per access technology pair define the behavior and interfaces forcommon core network components including the IP anchor, AAA, and HSS, aswell as the IP Multimedia Subsystem (IMS). This paper discusses theIntelligent Wireless Connection Manager (IWCM), a critical component of amobile device platform that enables seamless and transparent inter-technology mobility for user applications through IP session continuity,make-before-break handoff algorithms, and uniform quality of service (QoS)and policy management APIs. © 2011 Alcatel-Lucent.
expand inexpensive Wi-Fi* hotspots and 3G/4G fem-
tocell and picocell alternatives to Wi-Fi. One hundred
percent wireless service coverage with overlapping
areas of different access is becoming a reality.
User equipment (UE) vendors are responding
with a variety of multi-technology 3G, 4G, and Wi-Fi
mobile devices such as laptops/netbooks with built-in
or external USB-based modem cards, smartphones,
and iPADs*. Still, users expect not only to always stay
connected, but also to seamlessly maintain applica-
tion sessions while crossing the invisible boundaries of
different radio access technology (RAT) coverage
IntroductionFourth generation (4G) broadband wireless net-
work installations (e.g., Long Term Evolution [LTE]
and Worldwide Interoperability for Microwave Access
[WiMAX]) are expected to coexist for the duration of
a multi-technology transition period with a vast
embedded network of established high speed third
generation (3G) infrastructures including 3GPP2 High
Rate Packet Data (HRPD) and Evolved High Rate
Packet Data (eHRPD), 3GPP Wideband Code Division
Multiple Access (WCDMA) and Universal Mobile
Telecommunications System (UMTS). Meanwhile,
telecommunications and cable operators continue to
Bell Labs Technical Journal 15(4), 5–22 (2011) © 2011 Alcatel-Lucent. Published by Wiley Periodicals, Inc. Published online in Wiley Online Library (wileyonlinelibrary.com) • DOI: 10.1002/bltj.20469
6 Bell Labs Technical Journal DOI: 10.1002/bltj
areas. While a few applications (most notably e-mail,
web browsers, and certain file download programs)
are designed to maintain session continuity regard-
less of Internet Protocol (IP) address changes, many
applications such as Voice over Internet Protocol
(VoIP), video streaming, and secure virtual private
network (VPN) data sessions require underlying IP
session continuity as a prerequisite for this application
session continuity. In addition, real time applications
like video and VoIP require specific quality of service
(QoS) preservation. For the latter, packet loss, delay,
and general degradation of QoS conditions during
Panel 1. Abbreviations, Acronyms, and Terms
3G—Third generation3GPP—3rd Generation Partnership Project3GPP2—3rd Generation Partnership Project 24G—Fourth generationAAA—Authorization, authentication, and
accountingANDS—Access network discovery and selection ANDSF—ANDS functionAPI—Application programming interfaceCDMA—Code division multiple accessCM—Connection managerCMIP—Client MIPCoA—Care of addressDHCP—Dynamic Host Configuration ProtocolDL—DownlinkDR—Dual radioEAP—Extensible Authentication ProtocolEAP-AKA—EAP for UMTS Authentication and
Key AgreementeHRPD—Evolved high rate packet dataEPC—Evolved packet coreETSI—European Telecommunications Standards
InstituteE-UTRAN—Evolved UMTS terrestrial radio
access networkEV-DO—Evolution data optimizedFMCA—Fixed-Mobile Convergence AllianceGPRS—General packet radio serviceGTP—GPRS Tunneling ProtocolGUI—Graphical user interfaceHA—Home agentHRPD—High rate packet dataHSS—Home subscriber serverIMS—IP Multimedia ServiceIOT—Interoperability testIOTA—Integration of Two Access
(Technologies) Bell Labs ProjectIP—Internet ProtocolIWCM—Intelligent wireless connection
managerL2—Layer 2
L3—Layer 3LMA—Local mobility anchorLTE—Long Term EvolutionMAC—Medium access layerMIH—Media independent handoverMIHF—MIH functionMIP—Mobile IPMN—Mobile nodeNWG—Network Working GroupOS—Operating systemPCC—Policy and charging control PCRF—Policy and charging rules functionPDN-GW—Packet data network gatewayPMIP—Proxy MIPPoA—Point of attachmentQoS—Quality of serviceR8—Release 8RAN—Radio access networkRAT—Radio access technologyRF—Radio frequencySAP—Service access pointSAR—Specific absorption rateSP—Service providerSR—Single radioSSID—Service set identifierTLS—Transport Layer SecurityTTLS—Tunneled Transport Layer SecurityUE—User equipmentUL—UplinkUMTS—Universal Mobile Telecommunications
SystemUSB—Universal serial busVoIP—Voice over Internet ProtocolVPN—Virtual private networkWCDMA—Wideband code division multiple
accessWiMAX—Worldwide Interoperability for
Microwave AccessWISH—Wireless Interworking with Seamless
HandoversWLAN—Wireless local area network
DOI: 10.1002/bltj Bell Labs Technical Journal 7
inter-technology handover may lead to noticeable
quality degradation in the application session, at times
comparable to session loss. 3GPP standards specify a
requirement of sub-300 millisecond bearer traffic
interruption on LTE-eHRPD inter-technology hand-
over [14]. Seamless VoIP and high resolution video
may require bearer traffic interruptions of less than 50
milliseconds.
The problem of multi-technology interworking
with seamless handovers is challenging; there is no
generic solution that suits all combinations of wireless
access. Wireless standards (3GPP, 3GPP2, WiMAX) use
different network architectures, different authentica-
tion and authorization procedures and protocols, dif-
ferent IP mobility paradigms, different access network
discovery and selection methods, and different flavors
of quality of service flow establishment. Attempts to
define generic and universal mechanisms for inter-
working of wireless access technologies, such as media
independent handover (MIH) IEEE 802.21, are cur-
rently not embraced by the 3GPP, 3GPP2, and WiMAX
standards bodies. See also a different approach in [21].
4G standards have defined specific per access tech-
nologies pairs interworking architectures and proce-
dures to support inter-technology hand-offs with IP
session continuity [3, 19, 22, 24]. Consensus is grow-
ing that 3GPP Release 8 evolved packet core [2],
defined in [3] as the common architecture for LTE
interworking with other non-3GPP and 3GPP tech-
nologies, provides the path to commonality [3, 21,
22, 25].
Interworking architectures differ for single radio
(SR) and dual radio (DR) UEs. So called tightly cou-
pled interworking architecture with optimized inter-
technology handovers applies to both SR and DR
devices, and defines dedicated signaling and bearer
plane tunneling functions between the correspond-
ing radio access networks (RANs) of different access
technologies. Such specialized coupling allows for bet-
ter network side control over QoS bearer traffic char-
acteristics during inter-technology handovers,
conceptually somewhat similar to intra-technology
tunneling procedures. However, besides the associ-
ated network side complexity, SR interworking
imposes significant additional requirements on the
UEs in the handover preparation phase, to tunnel
medium access control (MAC) messages of the target
RAT while connected over the source RAT. This
requirement presents significant challenges for the UE
vendors, delaying the appearance of SR devices on
the market. In addition, many UE devices include
built-in Wi-Fi radio, resulting in the necessity of the
DR approach for interworking with 3G/4G.
This paper focuses on seamless interworking for
DR and multi-radio devices as a practical near-term
matter. For DR devices, the 3GPP, 3GPP2, and WiMAX
standards define loosely coupled interworking archi-
tecture with functional elements and procedures for
the network side support of IP session continuity on
inter-technology handovers [3, 10, 11, 14, 23–26].
However, in order to achieve a seamless user experi-
ence during inter-technology handovers, in addition
to the specified standards, some key specialized seam-
less inter-technology handover functions and algo-
rithms have to be implemented in the UE. The UE
operating systems (OS) currently on the market (e.g.,
laptop/netbook Windows*, Mac*, Android*, iPhone*,
Windows Mobile*, and embedded OSs used in hand-
sets or multi-technology USB dongles) do not have
native support for such functions.
We present a portable intelligent wireless con-
nection manager (IWCM) component add-on to the
UE mobile device platform, providing a critical miss-
ing component of uniform seamlessness for all applica-
tions. IWCM also serves as an application enablement
foundation, with technology-neutral application pro-
gramming interfaces (APIs) to wireless modem devices,
and an API mechanism to enforce a variety of service
provider (SP) interworking policies.
These research results are a product of multi-year
collaboration between the Bell Labs Wireless
Interworking with Seamless Handoffs (WISH) team
and Alcatel-Lucent product units on 4G, 3G, and Wi-
Fi interworking. Initial results received by the Bell
Labs (IOTA) project for 3G/Wi-Fi interworking were
published in [17, 18]; also see [16]. WISH IWCM
addresses specifics for seamless interworking in 4G
(LTE, WiMAX) access technologies, fully realizes a mix
of client-based and network-based mobility schemes,
and emphasizes the importance of access technology
neutral 802.21 media independent handover (MIH)-
like device adapter APIs and policy/QoS APIs.
8 Bell Labs Technical Journal DOI: 10.1002/bltj
We begin by discussing generic concepts of loosely
coupled interworking architecture for multi-radio
devices. We then focus on the UE role and functions
in the enablement of seamless handovers and present
details on the IWCM reference architecture, applica-
tions, and associated access technology neutral device
adapter APIs.
Interworking With Seamless Handovers for4G/3G/Wi-Fi Multi-Radio Devices: Basic LooselyCoupled Architecture
4G (LTE and WiMAX) interworking with 3G tech-
nologies and Wi-Fi has been extensively addressed by
the standards [3–6, 10, 11, 15, 23, 25, 26]. Figure 1illustrates the main components of a standards-
defined, loosely coupled interworking architecture
for multi-radio UE. Note that the so-called non-
optimized LTE-eHRPD interworking architecture might
utilize some components of tightly coupled inter-
working added to the basic loosely coupled premise,
e.g., measurements of target RAN available at the
source RAN [3, 9, 14].
Different access technology RANs are not aware
of each other; all intelligence of inter-technology han-
dovers is concentrated in the common core network
components and in the UE device.
The common core network components include
the IP mobility anchor, authentication and autho-
rization, the policy and charging control (PCC) func-
tion for enabling QoS-based IP services (e.g., IP
Multimedia Services [IMS]), and optional access net-
work discovery and selection function (ANDSF) for
assisting UE in discovering available networks.
Subsequent sections of this paper describe the
specialized intelligent connection manager function
in the UE device, a Bell Labs technology providing
Authentication/authorization
PCC
Commoncore
servicesApplicationIP services(e.g., IMS)
Radio accessnetwork
Technology A
Radio accessnetwork
Technology B
IPtransportnetwork
Wi-Fi accesswithout mobility
infrastructure
IP mobilityanchor
RAT A specificmobility tunneling
RAT B specificmobility tunneling
Multi-radioUE with ICM
Overlay CMIPtunnel between
UE and IPmobility anchor
ANDSFserver
3G, 4G or Wi-Fi†
with mobilityinfrastructure
3G—Third generation4G—Fourth generationANDSF—Access network discovery and selection functionCMIP—Client Mobile IPICM—Intelligent connection manager
IMS—IP Multimedia ServiceIP—Internet ProtocolPCC—Policy and charging controlRAT—Radio access technologyUE—User equipment
†Registered trademark of the Wi-Fi Alliance.
Figure 1.Components of loosely coupled interworking architecture for multi-radio UE.
DOI: 10.1002/bltj Bell Labs Technical Journal 9
the capability to perform make-before-break inter-
technology handover procedures in compliance with
a variety of technology-specific mobility standards,
which achieves seamlessness by minimizing inter-
technology handover latency, packet loss, delay, jitter,
reordering and QoS degradation.
In the remainder of this section we detail the
aforementioned common core network components,
together with access network discovery and selection
function and IEEE 802.21 considerations.
Common IP Mobility Anchor Function and IP SessionContinuity
4G (LTE, WiMAX) and 3G wireless standards
define RAT-specific IP mobility anchors for intra-
technology mobility. For each RAT, the corresponding
IP mobility anchor presents the registered UE device
as directly attached to the outside world, and main-
tains technology-specific mobility signaling and bearer
traffic tunneling to/from the serving RAN with the
current IP point of attachment (PoA) for the UE. As the
UE moves, its IP PoA may move as well (macro han-
dover). In conjunction with the handover, technology-
specific mobility re-registration of the new IP PoA
with the IP mobility anchor causes the latter to switch
the bearer tunnel to the new PoA. This allows the UE
to maintain IP session continuity with the same IP
address reachable via the IP mobility anchor.
Make-before-break inter-technology handover—
standard practice for multi-radio UE— refers to a pro-
cedure in which the UE continues an IP data session
over the old RAT while performing network entry
and over-the-air flow set up, and initiating IP con-
nectivity registration with the common IP mobility
anchor over the new RAT. We provide more details on
this procedure in the sections on common PCC and
seamless interworking algorithms below.
Interworking standards generally define the com-
mon IP mobility anchor function as a union of each
technology-specific IP mobility anchor. 3GPP, 3GPP2,
and WiMAX specifications devised significant differ-
ences in the choice of IP mobility anchors, mobility
modes (client-based versus network-based), and
mobility signaling protocols and tunneling mecha-
nisms. Examples are as follows:
• 3GPP2 [12] and WiMAX [24] define Mobile IP
(MIP)-based tunneling with MIP home agent
(HA) function as an IP mobility anchor (with
RAT-specific differences in MIP extensions and
authentication mechanism). 3GPP2 HRPD ini-
tially defined client-based mobility with the client
MIP (CMIP) function in the UE [12], and later
added network-based mobility with the proxy
MIP (PMIP) client function in the RAN [7].
WiMAX [24] allows CMIP, but existing deploy-
ments use PMIP; 3GPP2 eHRPD uses PMIP [8, 9].
• 3GPP general packet radio service (GPRS)-based
standards (including LTE and UMTS) define
network-based mobility with GPRS Tunneling
Protocol (GTP) or PMIP [2], with current deploy-
ments predominantly using GTP. Starting with
Release 8, 3GPP defines LTE-to-non-3GPP-based
interworking with the packet data network gate-
way (PDN-GW) as a common IP mobility anchor,
implementing local mobility anchor (LMA) MIP
HA functions to support CMIPv4 and PMIPv6-
based mobility anchoring from 3GPP2, WiMAX,
and wireless local area network (WLAN) Wi-Fi
access. This architecture allows for hybrid tun-
neling methods with the common IP anchor at
the PDN-GW: GTP over LTE or other 3GPP (e.g.,
UMTS) access (network based mobility), CMIP
over 3GPP2 HRPD, or optionally WLAN Wi-Fi:
PMIPv6 over 3GPP2 eHRPD and over WiMAX.
• Many Wi-Fi RAN installations do not have inher-
ent mobility agents. An upcoming standard [26]
defines new Wi-Fi RAN components with PMIP
network-based mobility and associated security
procedures for interworking with WiMAX. The
3GPP defines security, authentication, and autho-
rization procedures for Wi-Fi/3GPP interworking
in [4] and [6]. With mobility infrastructure pre-
sent in the Wi-Fi access network, network-based
mobility procedures apply. In the absence of
the mobility infrastructure in the Wi-Fi access
network (e.g., at a hotspot or coffee shop), a mul-
timode UE may use a CMIP overlay tunnel over
Wi-Fi access, as depicted in Figure 1, using CMIP
with a collocated care of address (CoA). Here, the
UE first obtains the local care of address and then
10 Bell Labs Technical Journal DOI: 10.1002/bltj
registers this care of address with the common
HA/LMA function to establish a MIP tunnel and
to obtain a home IP address for session mobility.
Security association procedures for the MIP ses-
sion are outlined in [3].
A multi-mode UE has to be aware of the mobility
mode (e.g., client-based with CMIP versus network-
based with GTP, PMIP, or other tunneling methods in
the network) supported by the access network for the
specific RAT, since the corresponding IP connectivity
establishment procedures differ significantly. With
network-based mobility, UE establishes IP connectiv-
ity as a non-mobile host on network entry, and the
network maintains IP session continuity on IP PoA
change. With client-based mobility, the UE explicitly
manages mobility by performing CMIP registration
with an IP mobility anchor on network entry; UE has
the responsibility to perform MIP session re-registration
any time the PoA changes. Inter-technology han-
dovers between RATs with heterogeneous mobility
modes (i.e., client-based and network-based mobi-
lity) require special functionality from the DR UE.
Common Authentication and Authorization FunctionThe common authentication and authorization
function allows the network to authenticate a multi-
mode UE device/user, and authorize IP sessions and
IP-based services via any available RAT.
3GPP, WiMAX, and 3GPP2 devised different
authentication/authorization procedures, methods, and
protocols requiring special reconciliatory interworking
standards for the corresponding access technology
pairs. 3GPP Release 8 introduced an evolved packet
core all-IP architecture with a 3GPP home subscriber
server (HSS) as a common home authentication/
authorization function, and a 3GPP AAA server as an
interworking conduit for other access technologies
utilizing AAA [2, 3]. WiMAX and 3GPP2 utilize tech-
nology-specific AAA servers for authorization,
authentication, and accounting (AAA), with inter-
working architecture defining common home AAA
with dual-mode access technology-specific interfaces
to the corresponding RANs and HA/LMA [11, 23].
Wi-Fi access also utilizes AAA [6]. Among technol-
ogy standards utilizing AAA, there are also differences
in Extensible Authentication Protocol (EAP) methods
used for authentication and keys distribution. WiMAX
defined EAP-Transport Layer Security (TLS) as
mandatory with EAP-Tunneled Transport Layer Security
(TTLS), and EAP for 3rd Generation Authentication and
Key Agreement (EAP-AKA) as optional. Meanwhile,
3GPP2 eHRPD uses EAP-AKA, which is different from
other Code Division Multiple Access (CDMA) technolo-
gies like 3G1x and HRPD/evolution data optimized
(EV-DO).
Multi-mode UE may need to maintain a consis-
tent technology-specific set of user/session creden-
tials for session continuity on inter-technology
handovers, and may need to transfer or map creden-
tials used for initial network entry via RAT A during
inter-technology handover to RAT B. Session re-
authentication on inter-technology handover using the
transferred credentials is performed via technology-
specific authentication/authorization methods.
Common PCCStarting with Release 8, 3GPP defined common
PCC architecture accepted by other wireless standards
[1]. This architecture defines the common policy and
charging rules function (PCRF) in the common core
network, which provides further access to common
IP-based application services, e.g., IMS. It allows sup-
porting dynamic RAT-specific over-the-air QoS flows
for IP-based applications, including VoIP, real time
video streaming, and real time gaming.
Preventing application session QoS degradation
during inter-technology handovers is challenging for a
number of reasons. Per technology pair mapping of
RAT-specific QoS flow profiles/templates is required;
for some RAT pairs it is already defined [14], for others
it might be non-trivial, especially when interworking
with Wi-Fi WLAN access. Different over-the-air flow
setup procedures are defined and implemented by vari-
ous standards. 3GPP2 HRPD deployments use predom-
inantly UE device-initiated QoS flow setups, though
network-initiated QoS is allowed [13]. The WiMAX
Forum Network Working Group (NWG) 1.5 has both
UE-initiated and network-initiated QoS flow setup, but
pre-NWG1.5 standard WiMAX deployments use only
a network-initiated QoS flow setup. 3GPP evolved
UMTS terrestrial radio access network (E-UTRAN)
allows both UE-initiated and network-initiated QoS
DOI: 10.1002/bltj Bell Labs Technical Journal 11
flow setup, with network-initiated QoS being preferred
and prioritized in deployments.
As a result, make-before-break procedures spe-
cific to each interworking technology pair are required
for seamless inter-technology handovers with QoS.
Common PCC needs to be access technology aware.
With homogeneous network-initiated QoS to network-
initiated QoS inter-technology handover, make-
before-break procedures for QoS flow establishment
over the target RAT are controlled by the PCRF, with
the UE device playing a passive role. With homoge-
neous UE-initiated QoS to UE-initiated QoS inter-
technology handover, make-before-break procedures
for QoS flow establishment over the target RAT are
controlled by the UE establishing QoS flows over the
target RAT prior to IP mobility switch, with PCRF
playing an authorization role. Heterogeneous inter-
technology handover procedures—those between
network-initiated QoS and UE-initiated QoS—are
more challenging. Here, make-before-break proce-
dures of setting QoS flows over the target RAT prior to
the IP mobility tunnel switch, and tearing down QoS
flows over the source RAT, need to be properly cor-
related between PCRF and UE.
802.21 ConsiderationsMedia independent handover is an IEEE 802.21
standards-based solution that defines access technology-
agnostic interworking procedures. The IEEE 802.21
working group, initially chaired by Ajay Rajkumar
from Bell Labs, began work in 2004, resulting in a
standard being approved in January 2009.
The goal of MIH is to facilitate handovers between
heterogeneous access networks through the definition
of a core set of services and components that would
exist within both UE and throughout the network.
MIH partitioned the services into three general cate-
gories: command, event, and information. Additionally,
MIH defines a protocol for communication between
MIH-capable nodes (e.g., mobile node [MN] to net-
work, or network to network) that allows for coordi-
nated efforts before, during, and after a handover.
An MIH-capable node contains an MIH function
(MIHF) that provides one or more of the core services
to the upper layers (MIH users), and expects a mini-
mum set of functions from the lower (link) layers.
As shown in Figure 2, MIH users gain access to MIH
services through an MIH service access point
(MIH_SAP), whereas MIHF gains access to lower layer
MIH users
MIH_SAP
MIH function (MIHF)(command, event and information services)
Remote(MIHF)
RAT 1
MIH_LINK_SAP
RAT 2
MIH_LINK_SAP
RAT 3
MIH_LINK_SAP
…RAT 4
MIH_LINK_SAP
MIH
_NET
_SA
P
MIH
_NET
_SA
P
UENetwork
MIH—Media independent handoverMIHF—MIH functionRAT—Radio access technology
SAP—Service access pointUE—User equipment
Figure 2.Media independent handover.
12 Bell Labs Technical Journal DOI: 10.1002/bltj
functions via an MIH link service access point
(MIH_LINK_SAP). When communicating with peer
MIHFs, it uses the MIH network service access point
(MIH_NET_SAP), which provides both layer 2 (L2) and
layer 3 (L3) transport. It is through the combination
of all of these service access points (SAPs) that MIH is
able to provide the three distinct services: command,
event, and information. (See [20] for further details).
The distributed nature of MIH requires a signifi-
cant initial investment in order to begin receiving
some of the benefits it provides. Now that MIH (or
more accurately IEEE 802.21) is an actual standard,
we should start seeing more effort in this area. In fact,
Bell Labs’ IEEE 802.21 MIH server recently completed
successful interoperability tests (IOTs) at a Fixed-
Mobile Convergence Alliance (FMCA) interoperabil-
ity event hosted and managed by the European
Telecommunications Standards Institute (ETSI).
The current 3GPP, 3GPP2, and WiMAX standards
do not include network side MIH components; UE
usage of MIH components is considered a vendor-
specific device implementation, outside the standards
scope. The WISH intelligent wireless connection
manager (IWCM) utilizes a set of MIH-like access
technology-neutral APIs, similar to a subset of MIH
command and event services related to control and
monitoring of state, behavior and presence of the
device, link, and network. See the IWCM section
below for details.
Access Network Discovery and Selection Function The process of active RAT re-selection at the UE
involves discovering available networks for each RAT,
evaluation of the potential handover candidates (by
taking into account operator policies, user preferences,
and signal conditions), and the inter-technology han-
dover trigger if the conditions are right.
Starting with Release 8, 3GPP defined the access
network discovery and selection function [3] in the
network to provide UE with the available access net-
work information and associated service provider poli-
cies. UE can obtain the IP address of the ANDSF as
part of host IP information on initial network entry.
Intelligent Wireless Connection ManagerThis section describes the IWCM, a portable mid-
dleware technology from the Bell Labs WISH project,
which provides a critical, seamless interworking add-on
component for a multi-radio mobile device platform.
When integrated with a UE device’s network pro-
tocol stack, the IWCM provides standards-compliant
(3GPP, 3GPP2, WiMAX, Wi-Fi) and service provider
policy-based functionality that enables seamless and
transparent session and service continuity on inter-
technology handovers to IP-based VoIP, video, and
data applications.
We introduce this section by describing the UE-
side procedures for seamless inter-technology han-
dover. Next, we present an overview of the IWCM
reference architecture, followed by a more detailed
description of its components.
The WISH IWCM prototype, available at Bell
Labs, is used for a variety of customer demonstrations,
trade shows, and ongoing research work.
Multi-Radio UE Procedures During Seamless Inter-Technology Handover
Networking standards outline the basics of multi-
radio UE behavior to support inter-technology handovers
with IP session continuity. In order to satisfy inter-
working requirements and service provider policies
for different access technologies, UE needs to support
a variety of mobility modes to maintain IP session
continuity on inter-RAT handovers. In addition, truly
seamless handover requires special functions and
operation procedures at the UE device to minimize
or eliminate packet data stream impairment, e.g.,
minimize handover latency, packet loss and reorder-
ing, and QoS degradation. This section highlights such
UE procedures for loosely coupled interworking archi-
tecture. Note that a tightly coupled architecture model
further increases the network role in seamless inter-
working, allowing implementation of some of the
described procedures (e.g., intelligent buffering) in
the network instead of in the UE.
Figure 3 illustrates inter-technology handoff pro-
cedures for dual radio UE.
• Step 1. At start time, access network discovery and
selection (ANDS) using pre-configured policies/
preferences together with dynamic signal quality
measurements yield selection of RAT A for initial
network entry.
• Steps 2–5. Outline procedures for network entry
with RAT A from the UE perspective.
DOI: 10.1002/bltj Bell Labs Technical Journal 13
• Step 6. As the UE moves across different access
coverage areas, the best access technology is being
re-evaluated. The ANDS algorithm in the UE
takes into account dynamic signal quality analy-
sis, together with analysis dynamically received
from the ANDSF server or preconfigured service
provider policies and user preferences. Special
considerations apply to handovers with QoS. As a
result, RAT re-selection occurs and B is chosen.
• Step 7. Authentication/authorization session cre-
dentials are transferred or mapped from the RAT
A device adapter to the RAT B device adapter for
network entry over the new access mode.
• Step 8. While the application session continues
over the active RAT A connection, a simultaneous
make-before-break over-the-air connection is
established over RAT B, using the credentials of
step 7. If RAT B supports it, the mobility mode
with a supported IP version and supported IP con-
nectivity establishment method is also negotiated
with RAN B. Alternatively, this information might
be received as part of the policies from the ANDSF
server over RAT A in step 6, or it may be statically
preconfigured.
• Step 9. IP mobility mode is selected for RAT B,
based upon information dynamically received in
step 6 and step 8, RAT B service provider policies
or preconfigured RAT B IP operation specifics.
• Step 10. Intelligent buffering starts for uplink (UL)
and downlink (DL) bearer traffic in conjunction
with make-before-break IP registration with the
common IP mobility anchor over the RAT B in
step 11. This is intended to minimize packet loss,
delay, jitter and reordering on data path switch.
• Step 11. Make-before-break IP connectivity estab-
lishment procedure is performed over RAT B,
ANDS—Access network discovery and selection algorithmANDSF—Access network discovery and selection functionCAN—Converged access networkIP—Internet ProtocolIWCM—Intelligent wireless connection manager
Commonauthentication/authorization
CommonIP mobility
anchor
CommonPCC
RANtechnology A
Dual radioUE withIWCM
1. Initial ANDS => RAT A
2. Establish over-the-air connection over RAT A. Authenticate.
3. Select IP mobilitymode for RAT A
4. Establish IP session and QoS flows over RAN A. Bidirectional bearer tunnel between RAN A and IP anchor.
11. Make-before-break. Establish IP session, and IP CAN QoS flows over RAN B. Bearer session switched to RAT B.
6. ANDS –reselection =>switch to RAT B
Get network information and possibly policy for IP mobilitymode for RAN B and additional credentials.
7. Transfer credentials
8. Make-before-break. Establish over-the-air connection over RAT B. Authenticate.
10. Start of bearer planebuffering + handover withQoS algorithm procedures
9. IP mobility mode forRAT B selected
12. Completion of bearerplane buffering
13. Uplink and downlink bearer traffic over RAT B, technology specific tunneling with IP anchor
RANtechnology B
ANDSFserver
5. Uplink and downlink bearer traffic over RAT A, technology specific tunneling with IP anchor.
PCC—Policy control and chargingQoS—Quality of serviceRAN—Radio access networkRAT—Radio access technologyUE—User equipment
Figure 3.Seamless inter-technology handover procedures for dual radio UE.
14 Bell Labs Technical Journal DOI: 10.1002/bltj
based upon the selections in step 9. IP level user/
session credentials from step 7 are used to authen-
ticate and authorize inter-technology IP session
transfer. The common IP mobility anchor, upon
receiving the binding registration (sent by the UE
over RAT B for client-based mobility, or by RAN B
for network-based mobility) re-affirms the same
IP address for the session and switches the data
path tunnel from the RAN A to the RAN B.
• Step 12. UE completes IP connectivity establish-
ment and buffering procedures started in step 10.
Bearer path under the virtual interface is switched
to RAT B.
• Step 13. IP and application sessions continue over
RAT B.
Seamless user experience on inter-technology
handover is created by a combination of make-before-
break handover, efficient network discovery and
selection, and buffering procedures in step 6 through
step 12.
IWCM Reference Architecture and FunctionalDescription
The IWCM function controls all UE aspects of RAT
selection and inter-technology handovers described
above. The IWCM behavior is controlled predomi-
nantly by service provider policies (dynamic or stati-
cally configured at the UE), with the addition of a
limited number of user preferences as allowed by ser-
vice provider policy. In this section, we present the
IWCM reference architecture followed by a more
detailed description of its significant components.
The reference architecture for WISH IWCM is
shown in Figure 4. It consists of two parts: a bearer
Applications
Connection manager
Protocol mobilityengine
Device adapter abstraction layer
WiMAXAuth
Bearer (data) plane
Policy/QoS agent
Policy clientGUI (optional)
Devices
LTE CDMA Wi-Fi† …
WiMAXLTE CDMA Wi-Fi …
Auth Auth Auth
Policy enforcement engine
Virtual adapter
TCP/IP stack
Control plane
User Kernel
Vendor/deviceproprietary APIs
Technology-neutral API
(e.g., 802.21)
QoS API
Buffering algorithmsTunneling *Standardization is highly
desirable
LTE—Long Term EvolutionQoS—Quality of serviceTCP—Transmission Control ProtocolWiMAX—Worldwide Interoperability for Microwave AccessWISH—Wireless Interworking with Seamless Handovers
Auth—AuthorizationAPI—Application program interfaceCDMA—Code division multiple accessGUI—Graphical user interfaceIP—Internet Protocol
†Registered trademark of the Wi-Fi Alliance.
Figure 4.Reference architecture for the WISH intelligent wireless connection manager.
DOI: 10.1002/bltj Bell Labs Technical Journal 15
(data) plane and a control plane. The top and bot-
tom halves of Figure 4 represent user and kernel
space, respectively. The IWCM bearer (data) plane is
implemented as a kernel driver; while its architec-
ture and algorithms are portable, the implementa-
tion is typically OS-specific and highly optimized for
performance. The IWCM control plane is imple-
mented as an application in the user space, and as
such, both its architecture and implementation are
highly portable.
Following is a brief functional description of
major components:
• Connection manager (CM) module. Controls access
link and connection state machines, network dis-
covery and selection, and network entry; man-
ages session continuity and handovers with
make-before-break procedures, and user and ses-
sion credentials transfer; controls policy enforce-
ment and real time policy related measurements
(e.g., signal quality); controls user plane and
seamless handover algorithms. The CM module
interfaces with the device adapters layer via
802.21-like technology neutral APIs, and receives
policy updates from external policy client appli-
cation via the policy/QoS agent abstraction layer.
• Protocol mobility engine. Implements a variety of
standards-defined RAT-specific IP level proce-
dures, including different mobility modes (client-
based and network-based) and associated IP
connectivity establishment methods.
• Policy/QoS agent. Implements an abstraction layer
for the policy agent API interfacing with the pol-
icy clients (e.g., ANDSF client or other service
provider policy client), and an abstraction layer
for application of the QoS API.
• Device adapter. Exposes technology-neutral
802.21-like APIs in the abstraction layer by hiding
vendor-specific device adapter driver APIs and
providing an example of the required plug-and-
play modem device support for seamless inter-
technology handovers.
• Bearer (data) plane. Manages a collection of physical
interfaces through a single virtual interface pre-
sented to the OS IP stack, directs bearer traffic to
and from the currently active device, implements a
policy enforcement engine controlled by the CM,
provides tunneling for client-based mobility with
collocated CoA, and implements UL and DL
buffering algorithms realization on handovers.
Additionally, it allows the protocol mobility engine
to bypass the OS IP stack, and send/receive MIP
and Dynamic Host Configuration Protocol (DHCP)
signaling messages over the new RAT, while con-
tinuing to direct application bearer traffic over the
old RAT, e.g., during make-before-break handovers.
• Graphical user interface (GUI). (Optional) graphical
user interface for configuring, monitoring, and/or
controlling the operation of the ICWM.
The IWCM design is flexible, extensible, and
highly configurable, allowing for the easy addition of
new access technologies with the ability to mix and
match devices, mobility protocols, and algorithms. The
list of technologies currently supported by the proto-
type WISH IWCM includes LTE, WiMAX, CDMA2000
(EV-DO/HRPD/eHRPD), WCDMA, and Wi-Fi.
In the following subsections, we detail the most
significant IWCM functions.
Policy control. The term policy refers to a broad set of
parameters controlling IWCM behavior. Such policies
are predominantly defined by service providers, and
can be classified into the broad categories of static (pre-
configured at the UE) or dynamic. As a matter of service
provider policy, some limited number of user prefer-
ences may also be allowed. The list of policy parameters
includes, but is not limited to, the following:
• Policies to configure only a subset of RAT adapters
to be controlled by IWCM under the virtual
adapter, resulting in a multi-homed UE interface
configuration, with seamless inter-technology
handoffs only between the subset of the inter-
faces grouped under the virtual adapter.
• Access network discovery and (re-) selection related
policies, including information available from the
ANDSF server, preconfigured network selection
preferences, handover trigger inputs (signal quality
range thresholds), and scanning parameters.
• Mobility mode and options to be used by the spe-
cific RAT (client based mobility with CMIP or net-
work based mobility) and IP version (IPv4, IPv6,
or dual stack IPv4v6).
• Optional (service provider allowed) user preferences
(e.g., manual handover with user intervention
16 Bell Labs Technical Journal DOI: 10.1002/bltj
versus automatic handover, interface priorities, and
profile of home Wi-Fi access point).
• Optional authentication credentials.
• Power and battery life management control poli-
cies impacting multiple radio operations.
• Overall control of radio frequency (RF) emissions
and transmit power for multiple radios.
• Technology-specific service level limitations, e.g.,
aggregated data limits for 3G access.
External policy client applications (e.g., ANDSF
client or other custom policy applications) can use
the policy agent API to communicate policies to the
IWCM. The policies can be statically preconfigured or
dynamically received from the network (specifics of
how the policy client obtains such policies are out-
side of the scope of this paper).
IWCM provides a policy enforcement function for
both the bearer and control plane, thus enabling ser-
vice providers to control multi-radio UE behavior.
Note that as a matter of policy, SPs may allow a user
to control certain predetermined sets of preferences
(e.g., a choice of manual versus automatic network
selection, or giving highest priority to home Wi-Fi or
other predefined Wi-Fi service set identifiers [SSIDs]
when in range).
CM and protocol mobility engine. Protocol mobility
engine operations are controlled by the CM, which
applies the relevant policies and provides credentials.
The mobility engine provides support for standards-
defined RAT-specific mobility modes (both client and
network based) with associated IPv4/IPv6 or dual
stack connectivity establishment methods.
An integrated MIP client function (client-based
mobility) supports a variety of MIP extensions specific
for WiMAX and 3GPP2 EV-DO/HRPD; for MIP version 4
(MIPv4), it also supports both foreign agent-based CoA
(as per WiMAX and 3GPP2) and collocated CoA (typi-
cal for Wi-Fi interworking) configurations. An inte-
grated DHCP client supports additional DHCP options
(e.g., carrying user identity or requesting an ANDSF
server IP address) necessary for interworking with net-
work-based mobility, as well as the acquisition of local
IP addresses for CMIP with collocated CoA.
CM also controls policy enforcement functions and
buffering algorithms in the user plane, in conjunction
with the make-before-break inter-technology han-
dover procedures.
Seamless interworking algorithms. A seamless
inter-technology handover experience is achieved via
a combination of several algorithms implemented in
the UE as listed below. Specific details of the algo-
rithms’ realization are outside the scope of this paper.
All algorithms are implemented in the CM, except for
intelligent buffering, which is implemented in the
bearer (data) plane and controlled by the CM.
• Enhanced network discovery and selection. The goals
of this algorithm include maintaining active RAT
with a signal quality within the policy-defined
boundaries; timely selection of a new RAT to mini-
mize the possibility of the currently active RAT
signal deteriorating below an acceptable level and
impacting the quality of application sessions;
avoiding a ping-pong effect of frequent back-and-
forth handovers between RATs at RAT coverage
boundaries; and enforcing both static and
dynamic SP policies. This algorithm makes use of
signal quality monitoring via an interface with
device adapters, in conjunction with associated
power and battery life policy.
• Make-before-break with dual radio. Assuming suffi-
cient access coverage areas overlap between the
new and the currently active RATs, a timely hand-
over trigger to the new RAT allows performing in
a make-before-break fashion network entry and
the establishment of IP connectivity over the new
RAT (using the second radio), while maintaining
a quality application session over the old RAT (via
the active radio). The IP mobility anchor switches
the data path tunnel from the old RAN to the new
RAN after processing the mobility binding regis-
tration message from the new RAN (e.g., the cor-
responding GTP message for LTE, or the
PMIP/CMIP message for non-3GPP technologies).
This reduces the possibility of a potential traffic
interruption gap up to the order of a mobility reg-
istration roundtrip delay between the UE and the
mobility anchor. Further improvements are pos-
sible using bicasting methods during handover
(e.g., MIP-defined simultaneous bindings with DL
bicasting) and by applying intelligent handover
DOI: 10.1002/bltj Bell Labs Technical Journal 17
buffering algorithms, controlled and timed by the
CM in juxtaposition with the make-before-break
procedures.
• Intelligent buffering during handover. When the UE
performs a make-before-break network entry and
the mobility anchor switches the data path tun-
nels from the old RAN to the new RAN, transient
UL packets over the old RAN can be lost, and
transient DL packets over the old RAN can arrive
out of order, especially on handover from 3G to
LTE, where the new leg is faster. Proper buffering
schemes in the UL and DL directions allow such
packet loss and reordering to be addressed.
Specialized per flow buffering may be necessary
for handovers with QoS. Additional DL data
stream synchronization algorithms may be
needed in the case of DL bicasting.
• Handover with QoS. The technology-neutral QoS
API exposed by the policy/QoS agent module pro-
vides an opportunity for higher layer applications
to request the creation/deletion of certain QoS
flows for those devices and RATs that support UE-
initiated QoS flow setup. One of the benefits of
providing this QoS API is that higher layer appli-
cations do not need to be concerned about the
specifics of how QoS flows are established for
each of the different access technologies. It has a
single, consistent view of how to request QoS-
specific flows. The CM, along with the device
adapter APIs, provide the necessary translation of
QoS parameters in order to create/delete a specific
QoS flow. The capability of the RAT to support
UE-initiated QoS flow setup is communicated as
a policy to the CM. As discussed above, support of
handovers with QoS between heterogeneous
access networks, where one of the networks has
a UE-initiated QoS flow setup and another has a
network-initiated QoS flow setup, requires coop-
eration between the PCC function and the UE (in
addition to a common PCC function in the net-
work). In the WISH IWCM architecture, such
cooperation is a function of the CM and device
adapter, using the device adapter API to retrieve
the currently configured QoS flows and to establish
specific QoS flows. For example, when handing
over from an access technology that supports
network-initiated QoS flow setup to an access
technology that supports a client-initiated QoS
flow setup, CM can (in a make-before-break fash-
ion) retrieve the current settings from the source
access technology device adapter, map the
retrieved QoS flow profile to the target access
technology, and then establish (as part of the
make-before-break handover procedure) the cor-
responding QoS flow on the target access tech-
nology.
• Power and battery life management. The power and
battery life management algorithm parameters
are provided to CM as a policy. The algorithm
allows battery usage to be minimized through a
variety of techniques, such as powering-up a sec-
ond radio only if the signal quality of the active
RAT radio degrades, or if location-based informa-
tion received from the ANDSF server indicates
the availability of the second RAT, controlling the
frequency of scans on the second radio.
• Controlling RF emissions and resolving specific absorp-
tion rate (SAR) issues for multi-radio UEs. When DR
UE has both radios on and it is located at a cell
boundary with poor signal quality for both RATs,
both radios begin transmitting at maximum power,
resulting in combined RF emissions exceeding
the limits specified by Federal Communications
Commission (FCC) regulations. The algorithm to
address this issue utilizes CM, minimizing instances
when both radios are turned on and controlling
the combined transmit power.
Device adapter APIs. The WISH IWCM defines a
set of technology-neutral IEEE 802.21 MIH-like APIs
between the CM and the device adapters, which are
used as a technology-independent abstraction layer
for controlling and monitoring the modem devices,
specifically in the context of seamless inter-technology
handovers. In turn, the device adapters layer trans-
lates the technology-independent APIs into access
technology-specific device adapter/driver APIs.
Currently, 3G/4G modem device vendors implement
proprietary vendor and technology device adapter/
driver APIs, resulting in the need for development
work in order to integrate each individual device with
API Description
(Synchronous command/responses)
Open/close API library API for opening and closing device control library.
Get/set device configuration Configures device and network related dynamic or static parameters,including network name, operator, packet data protocol context,operating modes or user profiles/credentials. Retrieves administrativestatus of the configured parameters.
Get device/link status Retrieves extended operational status for the device and link, e.g.,mode (idle/active), connection state, signal strength indicators, speed,and current configuration related to the connected network.
Get/set QoS Retrieves information on currently active QoS service flows (regardlessof whether network-initiated or UE-initiated QoS flow setup is usedby the RAT. Set up QoS flow when the RAT allows for UE-initiated QoSflow setup.
Scan for available networks Returns the list of available networks, together with thecorresponding NAP and NSP information, and signal strengthindicators.
Airlink up/down Set up or tear down of over the air connectivity with the selectedradio access network.
Get/set power mode API for power and battery management, e.g., putting the device intoor taking it out of power saving/idle/sleep mode, powering the deviceon or off, or setting Tx power levels.
Subscription This provides the capability to get/set dynamic session authenticationcredentials.
(Asynchronous notifications)
Mode change Provides a notification that a device experienced a mode change, e.g., transitioned to an idle, sleep, or active state.
Link up/down Provides a notification that an over the air connection was eitherestablished or torn down.
Link detected Provides a notification that a new network has been detected. Forexample, this may occur due to a change in position (i.e., a previouslyout-of-range access network is now detected) or perhaps by a changein signal conditions.
Link parameters report Provides a notification that handover-related thresholds have beencrossed. The thresholds for each of the corresponding parametersmust have been previously configured.
Intra-technology handoff Provides a notification of when an intra-technology handover beginsand/or ends. This information may be useful for delaying the decisionto perform an inter-technology handover as conditions may changesignificantly.
Scan complete Provides a notification that a previously initiated scan of availablenetworks has completed.
QoS setup complete Provides a notification that a previously initiated QoS set up hascompleted.
Airlink up/down complete Provides a notification that a previously initiated over the airconnection establishment or tear down procedure has completed.
Table I. Device adapter APIs.
API—Application programming interfaceNAP—Network access providerNSP—Network service providerQoS—Quality of service
RAT—Radio access technologyTx—TransmitterUE—User equipment
DOI: 10.1002/bltj Bell Labs Technical Journal 19
any type of connection manager. Standardization of
the device adapter APIs is required for true plug-and-
play functionality. Certain standardization activities
have been started by working groups in different stan-
dards bodies (e.g., WiMAX Forum Service Provider
Working Group), however the effort is still far from
complete.
Table I presents the set of device APIs identified
by the WISH IWCM specifically for multi-technology
interworking with seamless handoffs, including both
the synchronous command/response access and asyn-
chronous notifications. The former is used for config-
uration, monitoring, and control operations, whereas
the latter provides a performance-efficient mechanism
for monitoring real time events. Note that in the
absence of asynchronous notifications, implementa-
tions may choose the less performance-efficient
polling model.
ConclusionsIn this paper, we presented a new intelligent wire-
less connection manager (IWCM) functional compo-
nent add-on to the mobile device platform for
multi-radio UE devices. We discussed basic concepts of
the 3G/4G/Wi-Fi loosely coupled interworking archi-
tecture for multi-radio devices (single radio devices
are not expected in the near term), and demonstrated
that interworking with seamless handovers requires
new capabilities not present in current UEs. The addi-
tion of IWCM to the UE platform provides such capa-
bilities. We also presented a reference architecture
and high level functional description of the Bell Labs
WISH IWCM, optimized for seamless interworking
with 4G access technologies, e.g., LTE.
AcknowledgementsThe authors thank Peretz Feder for many fruitful
discussions and for his contribution to problem for-
mulations. The authors also wish to thank Scott
Miller, Milind Buddhikot, and Girish Chandranmenon
for making IOTA documentation and software avail-
able to the WISH project.
*TrademarksAndroid is a trademark of Google Inc.iPad, iPhone, and Mac are registered trademarks of Apple
Inc.
Wi-Fi is a registered trademark of the Wi-Fi Alliance.Windows and Windows Mobile are registered trade-
marks of Microsoft Corporation.
References[1] 3rd Generation Partnership Project, “Technical
Specification Group Services and SystemAspects, Policy and Charging ControlArchitecture,” 3GPP TS 23.203, Dec. 2009,�http://www.3gpp.org/ftp/Specs/html-info/23203.htm�.
[2] 3rd Generation Partnership Project, “TechnicalSpecification Group Services and SystemAspects, General Packet Radio Service (GPRS)Enhancements for Evolved Universal TerrestrialRadio Access Network (E-UTRAN) Access,”3GPP TS 23.401, Dec. 2009, �http://www.3gpp.org/ftp/Specs/html-info/23401.htm�.
[3] 3rd Generation Partnership Project, “TechnicalSpecification Group Services and SystemAspects, Architecture Enhancements for Non-3GPP Accesses,” 3GPP TS 23.402, Dec. 2009,�http://www.3gpp.org/ftp/Specs/html-info/23402.htm�.
[4] 3rd Generation Partnership Project, “TechnicalSpecification Group Core Network andTerminals, 3GPP System to Wireless Local AreaNetwork (WLAN) Interworking, WLAN UserEquipment (WLAN UE) to Network Protocols,Stage 3 (Release 9),” 3GPP TS 24.234, v9.0.0,Dec. 2009, �http://www.3gpp.org/ftp/Specs/html-info/24234.htm�.
[5] 3rd Generation Partnership Project, “TechnicalSpecification Group Core Network andTerminals, 3GPP Evolved Packet System,Optimized Handover Procedures and ProtocolsBetween E-UTRAN Access and cdma2000HRPD Access—Stage 3,” 3GPP TS 29.276, Dec.2009, �http://www.3gpp.org/ftp/Specs/html-info/29276.htm�.
[6] 3rd Generation Partnership Project, “TechnicalSpecification Group Service and SystemAspects, 3G Security, Wireless Local AreaNetwork (WLAN) Interworking Security(Release 9),” 3GPP TS 33.234, Oct. 2009,�http://www.3gpp.org/ftp/Specs/html-info/33234.htm�.
[7] 3rd Generation Partnership Project 2, “NetworkPMIP Support,” 3GPP2 X.S0054-220-A, v1.0, Aug. 2008, �http://www.3gpp2.org/Public _html/specs/tsgx.cfm�.
[8] 3rd Generation Partnership Project 2, “NetworkPMIP Support,” 3GPP2 X.S0061-0, v1.0, Dec. 5,
20 Bell Labs Technical Journal DOI: 10.1002/bltj
2008, �http://www.3gpp2.org/Public_html/specs/tsgx.cfm�.
[9] 3rd Generation Partnership Project 2,“Interoperability Specification (IOS) forEvolved High Rate Packet Data (eHRPD) RadioAccess Network Interfaces and Interworkingwith Enhanced Universal Terrestrial RadioAccess Network (E-UTRAN),” 3GPP2 A.S0022-0, v1.0, Mar. 2009, �http://www.3gpp2.org/Public_html/specs/tsga.cfm�.
[10] 3rd Generation Partnership Project 2,“HRPD/1XRTT and 3GPP E-UTRAN (LTE)Interworking and Inter-Technology Handoff:Stage 1 Requirements,” 3GPP2 S.R0129-A,v1.0, Apr. 2, 2009, �http://www.3gpp2.org/Public_html/specs/tsgs.cfm�.
[11] 3rd Generation Partnership Project 2,“Interoperability Specification (IOS) for HighRate Packet Data (HRPD) Radio AccessNetwork Interfaces and Interworking withWorld Interoperability for Microwave Access(WiMAX),” 3GPP2 A.S0023-0, v1.0, Apr. 2009,�http://www.3gpp2.org/Public_html/specs/tsga.cfm�.
[12] 3rd Generation Partnership Project 2,“cdma2000 Wireless IP Network Standard:Chapter 2—Simple IP and Mobile IP AccessServices,” 3GPP2 X.S0011-002-E, v1.0, Nov.2009, �http://www.3gpp2.org/Public_html/specs/tsgx.cfm�.
[13] 3rd Generation Partnership Project 2,“cdma2000 Wireless IP Network Standard:Chapter 4—Quality of Service and HeaderReduction,” 3GPP2 X.S0011-004-E, v1.0, Nov.2009, �http://www.3gpp2.org/Public_html/specs/tsgx.cfm�.
[14] 3rd Generation Partnership Project 2, “E-UTRAN—eHRPD Connectivity andInterworking: Core Network Aspects,” 3PGG2X.S0057-0, v2.0, Dec. 2009, �http://www.3gpp2.org/Public_html/specs/tsgx.cfm�.
[15] 3rd Generation Partnership Project 2, “WiMAX-HRPD Interworking: Core Network Aspects,”3GPP2 X.S0058-0, v2.0, June 2010,�http://www.3gpp2.org/Public_html/specs/tsgx.cfm�.
[16] D. Benenati, P. M. Feder, N. Y. Lee, S. Martin-Leon, and R. Shapira, “A Seamless Mobile VPNData Solution for CDMA2000, UMTS, andWLAN Users,” Bell Labs Tech. J., 7:2 (2002),143–165.
[17] M. Buddhikot, G. Chandranmenon, S. Han, Y.-W. Lee, S. Miller, and L. Salgarelli, “Design
and Implementation of a WLAN/CDMA2000Interworking Architecture,” IEEE Comm. Mag.,41:11 (2003), 90–100.
[18] M. Buddhikot, G. Chandranmenon, S. Han, Y.W. Lee, S. Miller, and L. Salgarelli, “Integrationof 802.11 and Third-Generation Wireless DataNetworks,” Proc. 22nd IEEE Internat. Conf. onComput. Commun. (INFOCOM ‘03) (SanFrancisco, CA, 2003), vol. 1, pp. 503–512.
[19] P. Feder, R. Isukapalli, and S. Mizikovsky,“WiMAX-EVDO Interworking Using Mobile IP,”IEEE Commun. Mag., 47:6 (2009), 122–131.
[20] Institute of Electrical and Electronics Engineers,“IEEE Standard for Local and MetropolitanArea Networks—Part 21: Media IndependentHandover Services,” IEEE 802.21-2008, Jan.21, 2009.
[21] R. Sigle, O. Blume, L. Ewe, and W. Wajda,“Multi-Radio Infrastructure for 4G,” Bell LabsTech. J., 13:4 (2009), 257–276.
[22] P. Taaghol, P. Feder, and R. Isukapalli, “MobileWiMAX Integration with 3GPP and 3GPP2Networks,” WiMAX Technology and NetworkEvolution (K. Etemad and M.-Y. Lai, eds.),John Wiley & Sons, Hoboken, NJ, 2010,Chapter 11.
[23] WiMAX Forum, Network Working Group,“WiMAX Forum Network Architecture—WiMAX—3GPP2 Interworking,” WMF-T37-004, May 2009.
[24] WiMAX Forum, Network Working Group,“Network Architecture—Release 1.5—Stage 3:Detailed Protocols and Procedures,” T33Protocol Specification, Sept. 2009.
[25] WiMAX Forum, Network Working Group,“WiMAX—3GPP EPS Interworking,” Draft T37-009-R016-v01, Oct. 2009.
[26] WiMAX Forum, Network Working Group,“WiMAX and WiFi Interworking,” WMF-T37-010-R016-v01, Sept. 2010.
(Manuscript approved September 2010)
EDWARD GRINSHPUN is team leader for the Alcatel-Lucent Bell Labs Wireless Internetworkingwith Seamless Handovers (WISH) project,which focuses on new methods,architecture, and algorithms to create andfurther enhance a seamless user experience
for handovers between 3G, 4G, and Wi-Fi wirelessaccess technologies. He is based in Murray Hill, NewJersey. Prior to joining Alcatel-Lucent, he helduniversity research positions in Israel and Canada.
DOI: 10.1002/bltj Bell Labs Technical Journal 21
Since joining Alcatel-Lucent almost 15 years ago, Dr. Grinshpun has worked as a technical team lead indevelopment and systems engineering for broadbandwireline and wireless products in the areas ofembedded systems efficiency, multi-service platforminfrastructure, high availability middleware, callprocessing, IP routing and multiprotocol labelswitching (MPLS), triple play content delivery, 4Gwireless networks architecture, and IP mobility andinterworking. He received the Bell Labs President’saward as a member of the WiMAX Technologies andInnovations project team, has authored over 20publications, and has served as an invited speaker at anumber of international conferences. He holds a Ph.D.in mathematical physics from Almaty University,Kazakhstan.
DAVID W. FAUCHER is a researcher in the Alcatel-Lucent Bell Labs Radio Access ResearchDepartment in Murray Hill, New Jersey. Hehas worked in a number of areas includingdigital transmission systems, securetelephones, IP security, and mobile and data
networking. He is currently working on wireless inter-technology with seamless handovers. Mr. Faucher has aB.S. in computer science from the University ofMassachusetts at Amherst, and an M.S. in computerengineering from the University of Lowell, Lowell,Massachusetts.
SAMEER SHARMA is research and development directorfor 4G wireless and LTE at Alcatel-LucentBell Labs in Murray Hill, New Jersey. Prior tojoining Alcatel-Lucent, he worked in variousmultinational, regional, and nationalcorporations in Singapore and India. He
received a B.E. in electronics engineering fromVisvesvaraya National Institute of Technology (VNIT),Nagpur, India, and an M.Tech. in computer science fromthe Indian Institute of Technology, Chennai. He has led and managed diverse global cross-functional projectsand teams developing complex new products andtechnologies, and has received numerous awards and recognitions during his academic life andprofessional career. He is a member of the Alcatel-Lucent Technical Academy, and a recipient of the BellLabs President’s Award for leading the WiMAXTechnologies and Innovations project. ◆