UDC 621.396.69:681.

32

A Study on Mobile IPv6 Based Mobility

Management Architecture

VTsuguo Kato VRyuichi Takechi VHideaki Ono

(Manuscript received January 19, 2001)

Mobile IPv6 is considered to be one of the key technologies for realizing Fixed Mobile
Convergence (FMC), which enables seamless communication between fixed and mo-
bile access networks. However, the original Mobile IPv6 does not support fast han-
dover, which is an essential function for mobile networks. One of reasons for this is
that only the home agent (HA) and correspondent nodes (CNs), which are not always
near to a mobile node (MN), manage an MN’s mobility in Mobile IPv6. To overcome this
problem, various hierarchical approaches have been proposed in the IETF. These ap-
proaches are good for achieving fast handover, however, they also sacrifice route op-
timization, which is one of the features of Mobile IPv6.
In this paper, we propose a novel mobility management architecture based on a hierar-
chical approach that can support fast handover without sacrificing route optimization.
We also analyze its effectiveness in regard to the number of handover-related messag-
es and the average handover time.

1. Introduction users who accessed the Internet though mobile

Lately, there has been a tremendous increase networks in Japan was already 12.7 million in
in the demand for mobile telecommunication net- 2000. Along with this expansion in demand for
works. For example, the number of subscribers to Internet access through wired and wireless means,
mobile telecommunication networks in Japan was Mobile IPv62),6),7) is considered to be one of the
38.2 million in 1998, 47.3 million in 1999, and key technologies for realizing Fixed Mobile
56.8 million in 2000.1) Convergence (FMC), which enables seamless com-
At the same time, the Internet is continuing munication between fixed and mobile access
its rapid expansion. The number of hosts in networks.3),9),10) However, the original Mobile IPv6
Japan was about 27 million in 1999 and will be could not achieve the fast handover required for
about 76.7 million by the year 2005. In addition mobile networks. One of the reasons for this is
to classical text-based services (e.g., NetNews and that the mobility management of Mobile IPv6 is
e-mail), the WWW provides users with easy net- handled only by the HA and CNs, which may be
work access and voice, image, and streaming far away from the MN. To overcome this problem,
transmissions. various hierarchical approaches have been pro-
In the upcoming decade, many people will posed by the IETF.4),5),8) These approaches are good
have personal computers and mobile terminals for achieving a fast handover, however, they sac-
with a Web browsing function, and these people rifice route optimization, which is one of the
will want to access multimedia information dis- features of Mobile IPv6. This inefficient routing
tributed via the Internet. In fact, the number of mechanism causes QoS deterioration and heavy

FUJITSU Sci. Tech. J.,37, 1, pp.65-71(June 2001) 65

T. Kato et al.: A Study on Mobile IPv6 Based Mobility Management Architecture

network congestion, especially in a public-scale (CNs), which can then send packets directly to the
network. MN. In this way, in Mobile IPv6, only the HA and
In this paper, we propose a novel mobility CNs manage an MN’s mobility. As a result, if the
management architecture based on a hierarchi- HA and/or CNs are far from the MN, even if the
cal approach that can support fast handover MN’s movement is small, BU messages travel
without sacrificing route optimization. In Section 2, across several IP networks, the latencies of which
we review the original Mobile IPv6 and Hierar- reduce the handover speed. Moreover route opti-
chical Mobile IPv6 approach and discuss some mization, which supports direct routing from CNs
issues of these existing mechanisms. In Section 3, to the MN, generates many BU messages and adds
we explain our proposed mobility management a significant extra load to the network.
architecture, and in Section 4 we evaluate its
effectiveness in regard to the number of handover- 2.2 Hierarchical Mobile IPv6
related messages and average handover time. To improve the performance of Mobile IPv6
in terms of handover speed and to reduce the num-
2. Existing Mechanism ber of BU messages sent to CNs and the HA, the
2.1 Mobile IPv6 overview IETF is currently discussing Hierarchical Mobile
The Mobile IPv6 protocol is currently being IPv6. Figure 2 shows the basic operation of Hi-
specified by the IETF IP Routing for Wireless/ erarchical Mobile IPv6. A new Mobile IPv6 node
Mobile Hosts working group. Figure 1 shows an called the mobility anchor point (MAP) is intro-
example operation of Mobile IPv6. Each time the duced. It simply provides an optional mobility
mobile node (MN) moves from one domain to an- management function that can be located at any
other, it gets a new care-of address (CoA). It then level in the hierarchy, starting from the access
registers its binding (association between its home router (AR) upwards. When an MN moves into a
address and CoA) with the home agent (HA) in MAP domain and attaches to an AR, the MN ob-
its home domain using Binding Update (BU) mes- tains a Regional Care-of address (RCoA) on the
sages. The HA records this binding in its Binding MAP’s domain and an on-link care-of address
Cache. The HA works as a proxy for the MN until (LCoA) from the AR. Then, the MN sends a BU
the MN’s binding entry expires or the MN returns message to the MAP. This BU message binds the
to its home domain. The HA intercepts any pack- RCoA and the LCoA. The MAP records this bind-
ets addressed to the MN’s home address and ing in its Binding Cache. The MN also sends BU
tunnels them to the MN’s CoA using IPv6 encap- messages to its HA and CNs. These BU messages
sulations. The MN also registers its binding by
sending BU messages to its correspondent nodes
CN
Carrier IP
network
HA
HA
Home Carrier IP Inefficient
AR Home
MN network network routing network

Carrier IP Internet q
MAP Carrier IP
network (Broker)
BU CN Internet network
Carrier IP BU
network AR (Broker)
CN
MN MN wBU Congestion

Figure 1 Figure 2
Mobile IPv6. Hierarchical Mobile IPv6.

66 FUJITSU Sci. Tech. J.,37, 1,(June 2001)

 T. Kato et al.: A Study on Mobile IPv6 Based Mobility Management Architecture

bind the home address of the MN and the RCoA. the MN sends a BU message to the THA
The MAP works like an HA. The MAP re- (Figure 3 s). This BU message binds the TCoA
ceives packets addressed to the MN’s RCoA from and the LCoA. The THA records this binding in
the HA or CNs. Packets are tunneled from the its Binding Cache. The MN also sends BU mes-
MAP to the MN’s LCoA using IPv6 encapsulations. sages to its HA and CNs if the MN has come from
The MN decapsulates the packets and processes another carrier IP network (Figure 3 d). These
them in the normal way. As a result, the MNs BU messages bind the home address of the MN
only need to send BU messages to the HA and and TCoA.
CNs when changing the MAP domain. Regional The THA works like an HA. The THA re-
BU (BU to the MAP) is sufficient for an MN with- ceives packets addressed to the MN’s TCoA from
in a MAP domain. In this way, Hierarchical Mobile the HA or CNs via G-EN (Figure 3 f). Packets
IPv6 enables an improvement in the handover are tunneled from the THA to the MN’s LCoA us-
speed of Mobile IPv6 and also enables a reduction ing IPv6 encapsulation (Figure 3 g). The MN
in the amount of BU messages sent to CNs and decapsulates the packets and processes them in
the HA. However, the essential point is that in the normal way. At the same time, when the
Hierarchical Mobile IPv6, packets tunnel from the G-EN receives packets addressed to the MN’s
MAP to the MN. This means that every packet to TCoA from the HA or CNs, the G-ENs send a Bind-
the MN travels via the MAP. If the MAP domain ing Request (BR) message to the MN’s TCoA
is very small, there may be no problems. Howev- (Figure 3 h). The THA intercepts this BR mes-
er, in a large-scale public network, this inefficient sage and registers a new binding which indicates
routing mechanism causes QoS deteriorations and the association between the home address of the
heavy node congestions, which is why Hierarchi- MN and the G-ENs’ addresses. The THA also
cal Mobile IPv6 approaches sacrifice route sends a BU message to the G-ENs (Figure 3 j).
optimization. This BU message binds the RCoA of the MN and
the LCoA. The G-EN records this binding in its
3. Proposed mobility Binding Cache. The G-ENs can then send pack-
management architecture ets directly to the MN (Figure 3 k). If the MN
To overcome the above problems of the exist- moves within the same carrier IP network, the
ing mobility management mechanisms, we BU messages the MN sends to its HA and CNs
propose a novel mobility management architec- are skipped.
ture based on a hierarchical approach which can In this architecture, a regional BU (BU to
support fast handover without sacrificing route
optimization. Figure 3 shows the basic concept
CN
of our mobility management architecture. Gate- Carrier IP
r network
way-Edge Nodes (G-ENs) and a Temporary Home G-EN
HA
Agent (THA) are introduced. The combination of
AR Carrier IP Home
G-ENs and the THA works as a MAP of the Hier- network r yBR uBU network
i THA
archical Mobile IPv6. The Binding Update (BU) uBU eBU
q t yBR
messages from the MN are almost the same as r G-EN Internet rCarrier IP
(Broker) network
those of Hierarchical Mobile IPv6. When an MN i
AR
moves into a THA domain and attaches to an AR, CN
wBU
MN eBU
the MN gets a Temporary Care-of address (TCoA)
on the THA’s domain and an on-link care-of ad- Figure 3
dress (LCoA) from the AR (Figure 3 a). Then, Proposed mobility management architecture.

FUJITSU Sci. Tech. J.,37, 1,(June 2001) 67

T. Kato et al.: A Study on Mobile IPv6 Based Mobility Management Architecture

the THA) is sufficient for any MN movement with- warding ratio to the HA and CNs of these BU mes-
in our proposed network. Once an MN roams in sages depends on the arrangement of service
our proposed network, the THA does not change. areas. Figure 5 shows an example of the rela-
When the THA receives a regional BU message tion between the arrangements of service areas
from the MN, the THA sends a BU message to the and the forwarding ratio to the HA and CNs of
G-ENs. The G-ENs can then send packets direct- BU messages when k equals 4. Figure 5 (a) shows
ly to the MN. The MN sends BU messages to the the best-arrangement case, R min, where the for-
HA and CNs only when an MN crosses the bound- warding ratio equals 7/12. Figure 5 (b) shows the
ary of our proposed network. In this way, our worst-arrangement case, R max, where the for-
proposed mobility management architecture can warding ratio equals 9/12. The forwarding ratios
reduce the amount of BU messages sent to the of the worst and best cases, are defined as follows:
HA and CNs and improve the handover speed of
Mobile IPv6 without sacrificing route optimization.
1 (1)
R max = (2 K + 1)
3k
4. Evaluation of the proposed 2x – 1 1 3x 2 – 3x + 0.5 – k
R min = – (2)
mobility management k 3k x–1
architecture
In this section, we theoretically evaluate the 1 4k – 1 .
where x = 1+
2 3
performance of our proposed mobility manage-
ment architecture. When evaluating the
performance of a mobility management scheme,
various factors have to be taken into account.
Here, we evaluate two important factors in our
proposal: 1) the efficiency of the hierarchical ap- AR AR
proach; that is, how much the hierarchical
approach reduces the number of messages, and
2) the efficiency of route optimization; that is, how
much the route optimization reduces the number
of transit nodes and backbone link bandwidth.
Figure 4
Service area model.
4.1 Efficiency of the hierarchical
approach. THA
In this chapter, we evaluate how much our
THA
hierarchical approach reduces the number of mes-
sages compared with the original Mobile IPv6. 2/3
1/2
Figure 4 shows the service area model of each
AR. We assumed that each AR has a hexagonal 1/2
5/6 2/3 2/3 5/6
2/3
service area and that each MN moves in a ran-
dom straight direction. We also assumed that, in
(a) Best case (b) Worst case
the original Mobile IPv6, the number of ARs in
Notes: The numbers in the hexagons show the forwarding
the network equals k. On the other hand, in our ratios to the HA and CNs.
hierarchical approach, only when an MN crosses
Figure 5
the boundary of our proposed network, the MN Example of relation between arrangements of service
sends BU messages to the HA and CNs. The for- areas and forwarding ratio to HA and CNs.

68 FUJITSU Sci. Tech. J.,37, 1,(June 2001)

 T. Kato et al.: A Study on Mobile IPv6 Based Mobility Management Architecture

Figure 6 plots R max and R min as k varies shown in Figure 7. The source terminal (HA and/
from 1 to 24. These plots show the effect of reduc- or CNs) and the destination terminal (MN) were
ing the number of messages sent to the HA and randomly selected from the leaf nodes (AR/G-EN).
CNs. For example, when the AR number k = 10, Based on this network model, we evaluated
the number of messages is reduced by about 30% the number of transit nodes between the source
in the worst arrangement and by about 60% in and destination. Compared with the convention-
the best arrangement. al mobility management mechanism, the proposed
mechanism achieves a 30% reduction in the num-
4.2 Efficiency of route optimization ber of transit nodes required when the distance
In this chapter, we evaluate how much the between the source and destination terminals
route optimization reduces the number of transit changes (Figure 8). This reduction leads to a re-
nodes and backbone link bandwidth compared duced propagation delay and provides users with
with the existing Hierarchical Mobile IPv6. The a high-quality service.
network model that was used for this analysis is We also found that as the distance between
the source and destination terminals increases,
the proposed mechanism achieves an increasing
advantage over the conventional mobility man-
1 agement in terms of the required bandwidth of
Forwarding ratio

0.8 the backbone link (Figure 9). For example, when

0.6 the distance between the source terminal and the
0.4
destination terminal is less than 4, the ratio of
bandwidth reduction is more than 60%. This re-
0.2
duction of bandwidth makes it possible to avoid
0
1 3 5 7 9 11 13 15 17 19 21 23 network congestion more efficiently.
k
R min R max

Figure 6
R max and R min.

100
Reduction in number of transit nodes (%)

90

80
AR/G-EN AR/G-EN
70
THA
AR/G-EN CN/HA 60
AR/G-EN

Backbone link 50

40

AR/G-EN 30
AR/G-EN
20
Di

AR/G-EN AR/G-EN
st

nc 10
a

e A
B e
Dis tanc
0 1 2 3 4 5 6 7
MN
Distance A (from source to destination)

Figure 7 Figure 8
Network model for route optimization analysis. Reduction in number of transit nodes.

FUJITSU Sci. Tech. J.,37, 1,(June 2001) 69

T. Kato et al.: A Study on Mobile IPv6 Based Mobility Management Architecture

100 number of transit nodes by more than 30% and

90 reduces the bandwidth of the backbone link by
80 more than 50% compared with the existing Hier-
Bandwidth reduction (%)

70
archical Mobile IPv6. This means that our
proposed mobility management architecture can
60
be used to create a large-scale mobile network that
50
provides seamless communication between fixed
40 and mobile access networks.
30

20 References
10 1) TCA: Preliminary FY1999 statistics on Tele-
communication Market in Japan. August,
0 1 2 3 4 5 6 7
2000.
Distance B (from THA to destination)
2) C. Perkins: IP Mobility Support. RFC2002,
Figure 9 October 1996.
Reduction in required bandwidth in backbone link.
3) T. Kato et al.: A Mobile Telecommunication
Network Architecture for the Multimedia
Age. ICC’99 S42.1, June 1999.
4) K. Malki and H. Soliman: Hierachical
5. Conclusion Mobile IPv4/v6 and Fast Handoffs.
The demand for Internet access through draft-elmalki-soliman-hmipv4v6-00.txt,
wired and wireless access is expanding, and March 2000.
Mobile IPv6 is considered to be one of the key tech- 5) G. Dommety: Local and Indirect Registration
nologies for realizing seamless communication for Anchoring Handoff. draft-dommety-
between fixed and mobile access networks. How- mobileip- anchor-handoff-00.txt, March 2000.
ever, the original Mobile IPv6 cannot support fast 6) S. Deering and R. Hinden: Internet Protocol,
handover and hierarchical approaches proposed Version 6 (IPv6) Specification. RFC2460,
in the IETF to support fast handover sacrifice December 1998.
route optimization, which is one of the features of 7) D. B. Johnson and C. Perkins: Mobility
Mobile IPv6. Support in IPv6. draft-ietf-mobileip-ipv6-12.txt,
In this paper, we proposed a novel mobility April 2000.
management architecture based on a hierarchi- 8) H. Soliman, C. Castellucia, K. El-Malki, and
cal approach that can support fast handover L. Bellier: Hierarchical MIPv6 mobility man-
without sacrificing route optimization. The HA agement. draft-ietf-mobileip-hmipv6-00.txt,
and CNs do not require any change from the orig- October, 2000.
inal Mobile IPv6. Our proposed scheme requires 9) 3GPP: General Packet Radio Service (GPRS)
only a few minor changes to the MN from the ex- Service description ; stage 2. TS23.060 v3.4.0,
isting hierarchical Mobile IPv6. Our scheme July 2000.
reduces the number of BU messages sent to the 10) A. Rouz et al.: Broadband Interworking Ar-
HA and CNs by between 20% and 60% compared chitecture for Future Mobile Multimedia
with the original Mobile IPv6. It also reduces the System. ICPWC99, Jaipur, India, Feb. 1999.

70 FUJITSU Sci. Tech. J.,37, 1,(June 2001)

 T. Kato et al.: A Study on Mobile IPv6 Based Mobility Management Architecture

Tsuguo Kato received the B.S. degree Hideaki Ono received the B.S. and M.S.
in Electronic Engineering from Waseda degrees in Electrical Engineering from
University, Tokyo, Japan in 1983. He Saga University, Saga, Japan in 1990
joined Fujitsu Ltd., Kawasaki, Japan in and 1992, respectively. He joined Fujitsu
1983 and was engaged in research and Laboratories Ltd., Kawasaki, Japan in
development of ATM switching systems 1992, where he has been engaged in
and Internet and mobile networks. research and development of commu-
He is currently a senior researcher at nication network systems. He is a mem-
the Network Systems Laboratories of ber of the Institute of Electronics,
Fujitsu Laboratories Ltd. Kawasaki, Information and Communication Engi-
Japan. He is a member of the Institute neers (IEICE) of Japan. He received
of Electronics, Information and Communication Engineers (IEICE) the Telecom Columbus Award from the Telecommunications Ad-
of Japan. vancement Foundation (TAF) in 1992.

Ryuichi Takechi received the B.S. de-

gree in Electrical Engineering from the
University of Tokyo, Tokyo, Japan in
1988. He joined Fujitsu Ltd., Kawasaki,
Japan in 1998, where he was engaged
in development of telephone switching
systems. In 1990, he moved to Fujitsu
Laboratories Ltd., Kawasaki, Japan,
where he has been engaged in
research and development of
telecommunication network systems.
He is a member of the Institute of Electronics, Information and
Communication Engineers (IEICE) of Japan.

FUJITSU Sci. Tech. J.,37, 1,(June 2001) 71