Você está na página 1de 7

Wireless Broadband Technologies

Hand-Off Evolution
with Multiple
Interfaces
Ching-Lun Lin and Chih-Hsiang Ho, Institute for Information Industry, Taiwan
Jen-Yi Pan, National Chung Cheng University, Taiwan

Multiple interface integrationparticularly during hand-offwill be


imperative in 4th-generation wireless networks. Interface scheduling
could resolve routing optimization, improve connection quality of
service, and save power, all at full bandwidth utilization, but new
issues such as simultaneous location must also be considered.

sers have experienced seamless handoff during cell phone calls when connected to single interfaces. In the future, however, seamless hand-off will
become commonplace even when devices are connected to multiple interfaces. The Ambient Networks project (www.ambient-networks.org) is developing a framework to let devices navigate complex network environments, allowing users access
to and control over communication services offered through multiple interfaces. A good multipleinterface management mechanism is clearly becoming essential.
A robust mechanism for handling multiple
interfaces can ease hand-off latency and packet
loss by using soft hand-off, which buffers essential
data for continuous transmission and prevents
22

IT Pro September/October 2008

interruptions in connections during hand-off.


The hand-off mechanism should manage multiple interfaces according to user requirements
with minimum control packets. Under a highdensity wireless network, control packet management with multiple interfaces becomes more
important than ever. The continuously heterogeneous hand-off scenario, which we call handaround, differs from hand-off in that it has highly
overlapped wireless signals, a complicated decision policy, and simultaneous location issues.
Basic hand-off decision policies and network
environment probing techniques make transmission through multiple interfaces more efficient.
In this article, we focus on multiple-interface
hand-off management mechanisms rather than
hand-off decision policies.

Published by the IEEE Computer Society

1520-9202/08/$25.00 2008 IEEE

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

The Evolution of Hand-Off


The term hand-off (also called hand-over) refers
to the process of transferring an ongoing call or
data session from one channel connected to the
core network to another (www.3gpp.org/specs/
specs.htm). In wireless IP networks, Mobile IPv4
(MIP)1 naturally defines the process of moving as
hand-off. Hand-off latency is the time between
the last connection and the next data transmission, but it also involves layer-two (L2) handoff delay (the interface-association time), layerthree (L3) hand-off delay (the IP registrationblocking period), and TCP/upper layer hand-off
delay (connection-reestablishment time).
As multiple interface technology has evolved,
so too has hand-offfrom homogeneous to
heterogeneous.1 Generally, we distinguish homogeneous and heterogeneous hand-off by the
types of interfaces involved. If we change the
communication interface after the hand-off
procedure, were doing a heterogeneous handoff. Users can simultaneously connect to different IP domains through multiple L2 interfaces
and synchronously transmit data on both links
in the overlapped wireless area. Hand-off in
this scenario is extremely complicated; we classify and analyze four hand-off methods for performing it based on different L2 and L3 models
in Table 1.
We can differentiate homogeneous from
heterogeneous hand-off by whether the communication device changes IP domains (either
remaining within Wi-Fi or changing from Wi-Fi
to 3G, for example). The terms intra and inter
also indicate whether users need to change IP addresses (for instance, moving from the 100.0.0.0
to the 200.0.0.0 IP domain). Table 1 highlights
four scenarios:

1.

2.

The user remains in a single IP domain but


changes to a different L2 base station (BS)
for the hand-off procedure. This is as simple
as a normal L2 hand-off procedure because
it occurs within one IP domain.
The user has multiple interfaces located
in different IP domains, and one interface
changes to a different L2 BS for the handoff procedure. Because were still using the
same interface, the user simply processes a
single-interface hand-off procedure, but we
need to perform a common L3 IP mobility

Table 1. Hand-off scenario classifications.


Domain
Network

Intra-IP
domain (layer 2)

Inter-IP
domain (layer 3)

Homogeneous
Heterogeneous

1. DL2
3. DIn + DL2

2. DL2 + DL3
4. DIn + DL2 + DL3

DL2: Layer 2 interface reassociation time.


DL3: Layer 3 MIPv4 / MIPv6 registration and IP reconfiguration

time.
DIn: Time it takes the system to start or switch different layer 2

interfaces.

3.

4.

procedure because the new L2 BS belongs to


a different IP domain.
The user remains in one IP domain and
changes to a different L2 interface for a
special purpose. The L2 wireless connections switch interfaces (for example, from
3G to Wi-Fi). Given that both interfaces
still belong to the same IP domain, we dont
need a new IP address to establish a new
connection.
The user employs multiple interfaces located
in different IP domains and changes to a different L2 interface for a special purpose. The
L2 wireless connections switch interfaces
and trigger the L3 IP mobility procedure because the new L3 IP domain is different.

In heterogeneous hand-off, we can perform a


soft hand-off by using another interface. Multiple connections help decrease hand-off latency
and data loss.
As Figure 1 illustrates, hand-around is a special case of heterogeneous hand-off. We define
it as an iterative hand-off procedure incorporated within a high-density network operation. 3
It takes place when a mobile station (MS) with
multiple interfaces moves into an area with
overlapping heterogeneous L2 wireless networks and different L3 IP domains. Operating
in a densely overlapped network, the MS always maintains at least one connected interface
while other interfaces process hand-offs. The
MS rarely disconnects due to lack of signals. In
hand-around, wireless coverage areas are highly overlapped, and two or more IP domains are
likely to overlap, which means that users can
synchronously transmit data through one or
more interfaces.

computer.org/ITPro

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

23

Wireless Broadband Technologies

Hand-around challenges some previously held


assumptions about hand-off processes:

Network construction. Wireless network construction has traditionally aimed to maximize


coverage while minimizing the number of BSs.
Both homogeneous and heterogeneous handoff act on this assumption. In hand-around
processing environments, such as airports and
conference rooms, heavy load-sharing or improved transmission quality should be primary
considerations in network construction rather
than just connection coverage.
Hand-off latency. As far as we know, most MSs
use only one network interface at a time. In
dense wireless environments, MSs can continuously receive wireless signals, but without
technical enhancements, users can experience
unacceptable hand-off latency. Soft hand-off
could eliminate latency and instability if we use
multiple interfaces. Devices can choose stable
links based on the probing information gathered by unused interfaces. Yet, decision making becomes difficult in high-density network
environments because the network conditions
change frequently, and latency can increase to
unacceptable levels. Hand-off latency occurs in

Domain 1

Domain 2

802.11 network
3G network
User move

heterogeneous hand-off, but the network complexity makes it even worse in hand-around.
Bandwidth exhaustion. In heterogeneous handoff, MSs can simultaneously receive packets
through multiple interfaces in an overlapped
area to avoid losing data. This method works
in small overlapped areas, but it quickly exhausts bandwidth in the quest for better data
stability and low hand-off latency. MSs concurrently use both interfaces for a long time
and exhaust redundant bandwidth, so even
though bandwidth exhaustion is generally
banished from hand-off decision policies, it
has a great influence on the decision to use
hand-around.
Hand-around is a continuous hand-off procedure used in high-density, highly complex heterogeneous wireless networks. It is an evolution
of hand-off, as Figure 2 illustrates. The MSs
automatic hand-off decision model assists in
eliminating hand-off latency and bandwidth exhaustion by considering real-time network environment conditions.

Current Hand-Off Approaches


A review of L2 and L3 hand-off mechanisms
demonstrates the evolution of multiple interfaces
technology.
Domain 3

Domain 4

Figure 1. Hand-around. This special case of heterogeneous hand-off occurs


when the user is in an environment with overlapping networks.

24

802.21
Figure 3 shows the IEEE
802.21 architecture, which
lies between L2 and L3 and
has a service access point
(SAP) connecting to L2 and
L3 individually. The purpose of 802.21 is to accelerate hand-off procedures
between different interfaces
(such as 802.11, 802.16, and
4G) and provide a handoff management entrance
(where users can execute
their own management
policies or mechanisms).4
The standard introduces
new functions to control
the low-layer network and
adds SAPs to provide infor-

IT Pro September/October 2008

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

Hand-off
Homogeneous
hand-off

Hand-around

Heterogeneous
hand-off

The event service initiates hand-off from the

Mobile IPv4
Simultaneous Binding for Mobile IP (SB-MIP) is
an MIP extension for multiple interfaces, defined
in RFC 3344.1 MSs can use simultaneous mobility binding with more than one network device to
avoid packet loss during the moving process. Figure 4 shows the SB-MIP procedure. MIP defines
an S bit for a binding update (BU) message that
indicates that simultaneous binding is required.
If the S bit is activated, the home agent (HA) will
reserve the previous care-of address (CoA) rather
than clear it. As the MS starts the hand-off procedure, the HA duplicates all the packets that were
originally forwarded to the MS and sends them
to all the IP addresses that the MS had previously
registered with the HA. As long as the MS retains at least one connection, it can continuously
receive packets without interruption during the
hand-off procedure.
Other researchers describe similar ways to decrease hand-off latency and data loss. Bechir Hamdaoui and Parameswaran Ramanathan discarded
data streaming duplication, instead using Reed-

Mediaindependent
hand-over
(MIH)
function
Event
service,
command
service,
information
service

Command
service
Event
service
Information
service

MIH users
Layer 3 or higher
mobility protocol

LLC_SAP
Command
service
Event
service
Information
service

MIH_LINK_SAP

Multiple interfaces can use these interactive services to communicate with each other. The lower
layer (that is, any layer below L3) uses events to
trigger special services, and the upper layer (L3
or higher) can perform the corresponding commands. Mobile devices use an interface provided
by the information service to speed up the handoff procedure.

Figure 2. Hand-off evolution. Hand-around is a


continuous hand-off procedure used in high-density,
highly complex heterogeneous wireless networks. We
believe it represents an evolution in hand-off.

MIH_SAP

lower layer by means of low-layer trigger events,


such as connection establishment or break and
signal decline.
The command service initiates hand-off from the
upper layer to control connectivityfor example, starting the hand-off process and switching interfaces.
The information service controls communications of basic static information, such as interface type, between the upper and lower layers.

Network 1
(for example
802.16)

LLC_SAP
MIH_LINK_SAP

mation to the upper layer. Both the user side and


the network side need to support 802.21 to communicate with each other.
The media-independent handover (MIH)
functiona major part of 802.21includes three
separate services:

Network 2
(for example
802.11)

Figure 3. IEEE 802.21 architecture. The 802.21 general


reference model introduces individual service access
points (SAPs) for different network layers. The lower
layer uses events to trigger special services, and the
upper layer can perform the corresponding commands.
Mobile devices use interface information provided
by the information service to speed up the hand-off
procedure.

Solomon codes for data-error correction.5 In their


approach, the MS achieves seamless hand-off by
simultaneously receiving data and error-checking
codes from different interfaces and fixing error
data at the same time. Using multiple connections
offers the advantage of reliably avoiding data loss,
but the complexity of combining duplicated data
streaming with redundant bandwidth exhaustion
also has its disadvantages.

computer.org/ITPro

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

25

Wireless Broadband Technologies

HA will logically appear as: HoA


BIDCoA.
Corresponded
An MS appends BIDs in the BU mesnode (CN)
sage to register multiple interfaces.
Foreign agent
Each registration has a unique BID.
Foreign agent
(FA)
(FA)
MCoA also defines a priority column in
the BU suboption for identifying userMobile station
desired interfaces to the HA. Thus, the
(MS)
HA can distinguish different L2 devices
used by a given MS and update specific
interface IP addresses according to the
user policy.
Binding update
The MCoA operates as shown in
Data transmission
Figure 5. The white circle in the midForeign
Foreign
dle of the figure indicates a wireless
network
network
network with large coverage (for example, 3G), and the small gray circles
Figure 4. Mobile IP with simultaneous binding. The HA duplicates
on both sides represent networks with
all the packets that were originally forwarded to the MS and sends
small coverage areas (such as WLANs).
them to all the IP addresses that the MS had previously registered
Although MCoA doesnt define how
with the HA.
many interfaces should be used in the
wireless overlapped area, we generally
assume a single transmission path and
All in foreign
Binding table
all movements occurring under foreign
network
(1) HoA, BID1, CoA1
networks.
(may update to CoA3)
Home
The MS sends a BU with a second
(20 HoA, BID2, CoA2
agent
interfaces BID and CoA2 as the MS
moves into the wireless overlapped area
for the first time. The MS then loses the
wireless connectivity associated with
Binding update
(use WLAN)
Binding update
CoA1 (represented in the middle of Fig(use 3G)
ure 5). Later, the MS receives a CoA3
from the first interface when moving
back into the wireless overlapped area.
The BU updates the IP address from
CoA1
CoA1 CoA2
CoA2
CoA2 CoA3
CoA3
CoA1 to CoA3, which can shrink the
growing binding table in the HA.
Use 3G
Use WLAN
Use WLAN
MCoA takes advantage of the HoA
BIDCoA mapping to increase flexibilFigure 5. Multiple care-of addresses registration. The binding
ity in using multiple interfaces. Conny
update updates the IP address from CoA1 to CoA3, which can shrink Larsson and his colleagues8 discuss the
the growing binding table in the HA.
quality-of-service (QoS) mechanism in
binding data streaming policy with difMobile IPv6
ferent BIDs, and Keigo Aso and Benjamin Koh
Multiple Care-of Addresses (MCoA) is a multifocus on BID mapping as they address data forple-interface management technology based on
warding issues that arise when a device connects
the Mobile IP version 6 (MIPv6) architecture;
to multiple interfaces.9 Little change is needed to
it defines and inserts unique binding identifithe original MIPv6 environment beyond practical
cation number (BID) columns in addition to
extensions. The disadvantage of using different
the original mapping from the home address
BIDs is the increasing number of control signals
(HoA) to the CoA.6,7 The binding table in the
resulting from the additional BID column.
Home agent
(HA)

26

IT Pro September/October 2008

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

Table 2. Comparison of Multiple-Interface Technologies.


Multiple-
interface
technology

OSI
network
layer

Interface
usage
(concurrent)

Requirement

Advantage

Drawback

802.21
Between One interface
Must write special Can display real-time Lack of mobile station

layer 2 and
service access point network conditions
information in core

layer 3
to individual
network; end-to-end

interface
devices both need

802.21 support
Simultaneous Layer 3 IPv4 Multiple
Must combine
Better transmission Redundant bandwidth
Binding for
interfaces
duplicated data
stability
exhaustion under
Mobile
streaming
high-density network
(SB-MIP)

environments
Multiple Care- Layer 3 IPv6 One or Adds binding Can flexibly register Almost none under
of-Address
multiple
identification IP addresses first and
high-density network
Registration
interfaces
column in control
use them later in the
environments
(MCoA)
message
core network

Problems and Open Issues


Table 2 lists various multiple-interface technologies. In L2 solution 802.21, the MIH function
lets the MS display real-time network conditions.
On the other hand, developing appropriate SAPs
could be difficult because of business considerations as well as the complexity of dealing with
multiple interfaces. If the mechanism is implemented in the IP layer, it will be transparent to
the link layer regardless of network heterogeneity. Thus, we consider L3 solutions to be more
suitable than L2 solutions.
MIP and MIPv6 mechanisms operate in L3
multiple interfaces. SB-MIP can simultaneously
receive multiple data types and offer better transmission stability than MIP. Yet, SB-MIP can lead
to bandwidth exhaustion in high-density networks, and it doesnt allow flexible use of multiple interfaces because of the simultaneous data
transmission.
MCoA places BID records in HA binding
tables, which enables the MS to register interfaces before using them. MCoA uses only one
transmission path in an overlapped area, which
can dramatically decrease bandwidth exhaustion. BID can also provide an L2 interface to
create flexible interface usage and support future technological advances, such as QoS and
routing optimization. Using BIDs requires additional memory space and slightly greater signal cost when using MCoA technology.
Several issues arise as researchers develop
technologies to handle multiple interfaces:

Routing optimization. Routing optimization has

traditionally focused on choosing the most


efficient data path, but that differs for each
interface. Packet forwarding by different interfaces could decrease transmission hops,
and the same idea could apply in general data
transmission.9
Connection QoS. Given that different transmission capabilities are represented in multipleinterface technologies, MSs can forward data
through different routing paths. Dividing data
streaming among different interfaces could increase transmission capabilities.
Power savings. An MS with multiple interfaces
can turn off specific interfaces to save power,
but this action neither saves the signal nor initiates power-saving mode. Nonetheless, using
scheduling interfaces could lead to dramatic
power savings.
Bandwidth utilization. Researchers must address
issues such as how to separate data streaming
in an HA and how to recombine data streams
in an MS. Possible directions include samedata transmission by all interfaces for better
stability or aggregating available bandwidth for
better transmission throughput.
Simultaneous location. Simultaneous location
occurs when an MS connects to a home network and a foreign network, using multiple
interfaces, at the same time. When an MS
moves back to its home network, however, it
can use only one interface for a mobile purpose (for example, to receive a cell phone

May/June 2008 IT Pro

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

27

Wireless Broadband Technologies

call). Researchers need to address this limitation when developing technology for using
multiple interfaces.
Connection-lost delay. As long as an MS remains
connected to a default router, there is no need
to execute a hand-off procedure. If an MS connects to a default router through two interfaces
and loses one connection when the MS moves
away from a router, the HA will continue using
the incorrect MS IP address until the L2 interface detects the lost connection. During that
time, the user could miss calls. Moreover, if
the L2 interface is inadvertently placed in sleep
mode, hand-off latency will increase.
Interface selection. Major issues in interface
selection include which to use and when to
change. The appropriate choice depends on
signal strength, stability, efficiency, and power
consumption.
We should take a different view of interfaces
to resolve traditional network problems. As
multiple-interface technology progresses, it simultaneously solves existing problems and creates new issues.

s they continue to develop, technologies


using multiple interfaces will be expected to resolve traditional network problems as well as address new issues that arise.
Interface scheduling is one possible solution
to resolve routing optimization, connection
QoS, power savings, and bandwidth utilization.
Under the upcoming 4G network, devices will
need to be equipped with appropriate technology to manage multiple interfaces. For example,
researchers are working on interface scheduling, using layer 7 to control interface usage for
better performance.

References
1.
2.

3.

28

C. Perkins, IP Mobility Support for IPv4, IETF RFC


3344, Aug. 2002; www.ietf.org/rfc/rfc3344.txt.
N. Shenoy and R. Montalvo, A Framework for
Seamless Roaming across Cellular and Wireless Local Area Networks, IEEE Wireless Comm., vol. 12, no.
3, 2005, pp. 5057.
J.-L. Lin and J.-Y. Pan, Hand-Around: A Hand-Off
Evolution with Monami6, Proc. 3rd IEEE Intl Conf.

4.

5.

6.

7.

8.

9.

Wireless Comm., Networking, and Mobile Computing


(WiCOM 07), IEEE CS Press, 2007, pp. 17751778.
IEEE 802.21/D9.0, Draft IEEE Standard for Local and
Metropolitan Area Networks: Media Independent Handover
Services, IEEE, 2008.
B. Hamdaoui and P. Ramanathan, A Network-Layer
Soft Hand-Off Approach for Mobile Wireless IPBased Systems, IEEE J. Selected Areas in Comm., vol.
22, no. 4, 2004, pp. 630642.
R. Wakikawa, T. Ernst, and K. Nagami, Multiple
Care-of Addresses Registration, IETF Internet draft,
May 2008; work in progress.
D. Johnson, C. Perkins, and J. Arkko, Mobility Support in IPv6, IETF RFC 3775, June 2004; www.ietf.
org/rfc/rfc3775.txt.
C. Larsson et al., A Filter Rule Mechanism for Multiaccess Mobile IPv6, IETF Internet draft, Mar. 2007;
work in progress.
K. Aso and B. Koh, Multiple Forwarding Destinations Notification, IETF Internet draft, Jan. 2006;
work in progress.

Ching-Lun Lin is an associate engineer with the Institute for Information Industry (III), Taipei, Taiwan,
R.O.C. His research interests include WiMAX and mobility management. Lin has a BS in computer science and an
MS in communication engineering from National Chung
Cheng University, Min-Hsiung, Taiwan. Contact him at
jerry0422@nmi.iii.org.tw.

Chih-Hsiang Ho is a section manager of the WiMAX


Technology Center at III. His research interests include
mobile and wireless networking. Ho has a BS in aerospace
engineering from Tamkang University, Taipei, and an MS
in applied mechanics from National Taiwan University,
Taipei. Ho is now a PhD student at National Taiwan University, Taipei. Contact him at andrew@nmi.iii.org.tw.

Jen-Yi Pan is an assistant professor in the Department


of Communications Engineering and Center for Telecommunication Research, National Chung Cheng University,
Chaiyi, Taiwan. His research interests include performance
evaluation of medium access control and mobility management. Pan has a BS and a PhD in computer science from
National Tsing-Hua University, Hsinchu, Taiwan. He
is a member of the ACM and the IEEE. Contact him at
jypan@ccu.edu.tw.

IT Pro September/October 2008

Authorized licensed use limited to: MOTILAL NEHRU NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on August 12, 2009 at 01:53 from IEEE Xplore. Restrictions apply.

Você também pode gostar