Mobility Management Approaches for Mobile

IP Networks: Performance Comparison

and Use Recommendations
Nadjia Kara, Member, IEEE
AbstractIn wireless networks, efficient management of mobility is a crucial issue to support mobile users. The Mobile Internet
Protocol (MIP) has been proposed to support global mobility in IP networks. Several mobility management strategies have been
proposed which aim reducing the signaling traffic related to the Mobile Terminals (MTs) registration with the Home Agents (HAs)
whenever their Care-of-Addresses (CoAs) change. They use different Foreign Agents (FAs) and Gateway FAs (GFAs) hierarchies to
concentrate the registration processes. For high-mobility MTs, the Hierarchical MIP (HMIP) and Dynamic HMIP (DHMIP) strategies
localize the registration in FAs and GFAs, yielding to high-mobility signaling. The Multicast HMIP strategy limits the registration
processes in the GFAs. For high-mobility MTs, it provides lowest mobility signaling delay compared to the HMIP and DHMIP
approaches. However, it is resource consuming strategy unless for frequent MT mobility. Hence, we propose an analytic model to
evaluate the mean signaling delay and the mean bandwidth per call according to the type of MT mobility. In our analysis, the MHMIP
outperforms the DHMIP and MIP strategies in almost all the studied cases. The main contribution of this paper is the analytic model
that allows the mobility management approaches performance evaluation.
Index TermsMobile IP, mobility approach, performance evaluation.

1 INTRODUCTION
I
P multimedia applications are becoming popular in the
packet-based wireless networks. The integration of these
applications in wireless networks requires the support of
seamless terminal mobility. Mobile IP (MIP) has been
proposed by the Internet Engineering Task Force (IETF) to
provide global mobility in IP networks [1]. It allows
maintaining mobile terminals ongoing communications
while moving through IP network [1], [2].
In the MIP protocol, Mobile Terminal (MT) registers with
its home network from which it gets a permanent address
(home address). This address is stored in the Home Agent
(HA). It is used for identification and routing purpose. If
MT moves outside the home network visiting a foreign
network, it maintains its home address and obtains a new
one from the Foreign Agent (FA). This Foreign address is
called Care-of-Address (CoA). To allow continuity of
ongoing communications between the MT and a remote
end point, the MT shall inform the HA of its current
location when it moves outside the home network. The HA
delivers to MT the intercepted packets by tunneling them to
the MTs current point of attachment.
IP mobility in wireless networks can be classified into
macro- and micromobility. The macromobility is the MT
mobility through different administration domains. The
micromobility is the MT movements through different
subnets belonging to a single network domain. For
micromobility where the MT movement is frequent, the
MIP concept is not suitable and needs to be improved [3].
Indeed, the processing overhead related to location update
could be high specifically under high number of MTs and
when MTs are distant from the HAs yielding to high-
mobility signaling delay [4].
Hierarchical Mobile IP (HMIP) has been proposed to
reduce the number of location updates to HA and the
signaling latency when an MT moves from one subnet to
another [5], [6]. In this mobility scheme, FAs and Gateway
FAs (GFAs) are organized into a hierarchy. When an MT
changes FA within the same regional network, it updates its
CoA by performing a regional registration to the GFA.
When an MT moves to another regional network, it
performs a home registration with its HA using a publicly
routable address of GFA. The packets intercepted by the
HA are tunneled to a new GFA to which the MT is
belonging (e.g., G1
2
following MT handoff from 1
3
to
1
5
in Fig. 1). The GFA checks its visitor list and forwards
the packets to the FA of the MT (1
5
in Fig. 1). This
regional registration is sensitive to the GFAs failure because
of the centralized system architecture [7], [8]. Moreover, a
high traffic load on GFAs and frequent mobility between
regional networks degrade the mobility scheme perfor-
mance [4]. In order to reduce the signaling load for
interregional networks, mobility dynamic location manage-
ment approaches for MIP have been proposed: A Hier-
archical Distributed Dynamic Mobile IP (HDDMIP) and
Dynamic Hierarchical Mobile IP (DHMIP).
In the HDDMIP approach, each FA can act either as an
FA or GFA according to the user mobility. The traffic load
1312 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
. The author is with INRS Energy, Materials et Telecommunications (INRS-
EMT), Place Bonaventure, 800, de La Gauchetie`re West, Gate North-West,
Suite 6900, Montreal, Quebec, Canada, H5A 1K6.
E-mail: kara@emt.inrs.ca.
Manuscript received 21 Sept. 2007; revised 4 Jan. 2009; accepted 21 Jan. 2009;
published online 6 Feb. 2009.
For information on obtaining reprints of this article, please send e-mail to:
tmc@computer.org, and reference IEEECS Log Number TMC-2007-09-0290.
Digital Object Identifier no. 10.1109/TMC.2009.36.
1536-1233/09/$25.00 2009 IEEE Published by the IEEE CS, CASS, ComSoc, IES, & SPS
in a regional network is distributed among the FAs. The
number of FAs attached to a GFA is adjusted for each MT.
Thus, the regional network boundary varies for each MT.
This number is computed according to the MT mobility
characteristics and the incoming packet arrival rate. This
number is adjustable from time to time according to the
variation of the mobility and the packet arrival rate for each
MT. In [9] and [10], analytic models are proposed to
compute this number such as the total signaling traffic for
location update and packet delivery is transferred with
minimal network resource and low delay, respectively.
Nevertheless, this approach requires that each FA is able to
act as an FA and a GFA. Moreover, it adds processing load
on the MT to estimate the average packet arrival rate and
the subnet residence time. Hence, the main advantage of
this approach is the system robustness enhancement since
the GFA failure affects only the packets routing to MTs
belonging to this GFA. The disadvantages are the system
infrastructure and MTs costs which could be high.
The DHMIP approach has been proposed to reduce the
location update messages to the HA by registering the new
CoA to the previous FA and building a hierarchy of FAs .
Hence, the users packets are intercepted and tunneled
along the FAs hierarchy to the MT. The hierarchy level
numbers are dynamically adjusted based on mobile users
mobility and traffic load information. Fig. 2 illustrates an
example of DHMIP approach with a maximum of hierarchy
level number equal to 3. When MT is attached to 1
2
, 1
3
,
1
5
, or 1
6
, the CoA update is sent to the previous FAs. If
the MT becomes attached to 1
4
the level number reach the
threshold and the MT will set up a new hierarchy. The MT
registers its new CoA directly to the HA. In this approach,
the location update to the new FA, which is close to the
previous FAs, could be less expensive than that to the HA.
In [11], authors propose an analytic performance model to
evaluate the signaling transmission, the packet delivery, and
the total costs of HMIP, HDDMIP, and DHMIP mobility
approaches using a one-dimensional random walk model.
The performance analysis shows that the DHMIP scheme
outperforms compared to the HMIP and HDDMIP ones.
Despite that, the DHMIP approach still requires the new
location update and packet route processing in FAs
belonging to the hierarchy increasing the mobility signaling
and packet delivery delay. Moreover, the path extension
through the FAs hierarchy increases the network resources
used for packet delivery and location update signaling for
an ongoing communication.
In [12], another inter-FAs tunneling approach has been
proposed to optimize the route between the remote end
point and the MT. This approach enables remote end point
to get the CoA associated to the MT and to use it to reach
the MT through the foreigner network without passing
through the home network. When the MT moves from one
foreigner network to another, it communicates its new CoA
to its previous FA through its new FA. The previous FA
tunnels the received traffic from the remote end point to the
MTs new location. At the same time, it sends a message to
the HA requesting that the remote end point be notified of
the MTs new CoA. Upon receiving this new CoA, the
remote end point uses it to reach the MT through the new
foreigner network without passing through its previous
foreigner network. This approach requires to restore an
optimized route after each CoA change. It aims to transfer
packets through the resulting route with smaller delay than
that experienced when these packets transit through the
home network. However, this may not be always the case,
and such performance will depend on the route optimiza-
tion mechanism used and a set of influencing factors such
as remote end point to FAs distance, the loads of the
networks the optimized route should pass through, and the
MT inter-FAs mobility frequency. Such analysis is needed
to compare this approach with the existing ones, but it is out
of the scope of this paper.
Another alternative that reduce the signaling load in
Mobile IP network is to use a multicast-based mobility
approaches. These approaches have been proposed to
reduce the mobility signaling delay by setting a multicast
group (see Section 2). The MTs address update processes
are concentrated into the multicast network nodes (e.g.,
routers). They are reachable under these multicast group
addresses. However, these approaches could be resource
consuming except for next-generation IP-based radio access
technologies such as 3rd Generation Partnership Program
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1313
Fig. 1. MIP and DHMIP mobility approaches.
Fig. 2. DHMIP mobility approaches.
(3GPP) and 3GPP2 future cellular communication system
called Long Term Evolution (LTE) [13], [14]. In LTE
systems, where small cells deployment is expected, MT
with high mobility will be able to access different wireless
networks frequently yielding to increase traffic overhead
due to MIP signaling and tunneling. This signaling includes
not only location update signaling but also security
association signaling required for MIP support [14], [15].
HAs could be signaling traffic bottleneck for such future
mobile networks with high-mobility MTs. Hence, MHMIP
mobility approach is proposed to reduce the signaling delay
using multicast groups. The MT with high mobility could
reuse the multicast resources for signaling and packet
delivery for several handoff events that occur during its call
holding time. From that we expect that the resource usage is
no greater than that of the DHMIP mobility approach.
Hence, we propose to compute the mean bandwidth per
call and the mean handoff delay per call used for signaling
and packet delivery according to the MT mobility and call
holding time duration, and to compare the performance of a
Multicast Hierarchical Mobile IP approach (MHMIP) with
those of the DHMIP and MIP mobility strategies. We derive
a set of recommendations for the usage of these mobility
management approaches according to the MTs mobility.
The main contribution of this paper is the analytic model
that allows performance evaluation of three mobility
management approaches.
This paper is organized as follows: Section 2 discusses
the multicast-based mobility approaches. Sections 3 and 4
present the analytic model and the numerical results,
respectively. Section 5 gives the conclusions.
2 MULTICAST-BASED MOBILITY APPROACHES
2.1 Overview
The multicast has been proposed to be used for mobility
support and specifically in wireless networks with small
radio cells and high mobility of MTs. Several multicast-
based mobility approaches have been proposed. They can
be classified into multicast-based mobility in connection-
oriented and connection-less networks. For connection-
oriented networks, Acampora and Naghshineh propose a
virtual tree concept, where a multicast connection tree is
preestablished. This tree is a collection of radio base stations
and ATM network switches connected to the trees root. The
signaling delay is limited to the activation and deactivation
of preestablished branch in the tree [16].
For Connection-less network, Seshan, in [17], proposes to
apply a multicast to Mobile IP to reduce the handoff delay.
The HA encapsulates the intercepted packets into multicast
packets and sends them to the targeted MT over multiple
FAs. In [18], Ghai and Singh propose to divide the wireless
network into regions controlled by a supervisor host. Each
region includes groups of cells such as each cell may be part
of several of these groups. A unique IP multicast ID is
assigned to each of these groups. In [19], authors extend this
work by considering multiple wireless networks and cases
where mobile device is not able to use channel character-
istics to trigger handoffs due to the frequent network
interface change.
Different Mobile IP multicast protocols have been
proposed. In [20], Mobility Supporting Agents (MSA)-based
architecture has been proposed using IGMPv2 and PIM SM
IP multicast protocols. In [21], an Core Based Trees (CBT)-
based multicast mobile IP approach has been proposed for
micromobility. In [22], authors propose a set of multicast
mobility protocols called Candidate Access Router set (CAR-
set). The performance of multicast mobility approaches has
been evaluated through simulation or through analytic
models [22], [23]. In [22], a set of performance metrics (such
as handoff delay, packet loss, and bandwidth overhead due
to handoff) have been identified and evaluated for multicast
mobility approaches that have been simulated using NS2
network simulator. In [23], a software platform, set up
testbeds, has been used to analyze multicast mobility
protocols in terms of handoff delay, packets losses and
duplications, and relative TCP throughput. There is a large
number of multicast approaches that could be used to
implement mobility into MIP networks. The analysis of these
approaches andtheir design is not the focus of this paper. We
refer the reader to [23], where four case studies for multicast-
based mobility are presented based on different multicast
service models and protocols. In this paper, we focus on
usage of the multicast hierarchical architecture for IP
mobility support and its performance in terms of bandwidth
usage and handoff delay. The example used in this paper of
such architecture is given in Section 2.2.
2.2 Multicast Hierarchical Mobile IP
In this approach, we propose to build hierarchical multicast
groups. In each group, FAs are connected to each other
through a GFA. A set of GFAs are connected to an HA.
When an MT moves through FAs belonging to the same
group, the GFA of this group multicasts the received packet
(coming from the HA) to the MT. When the MT moves
outside a group, the new CoA is registered to the GFA of
the new group to which the MT is currently belonging. This
GFA sends this CoA to the HA. This latest tunnels the
packet to the new GFA which will multicast the received
packets within the new FAs group. This approach reduces
the frequency of the location update to the HA. This update
is performed every inter-GFAs mobility rather than every
inter-FAs mobility limiting the location update processing
only at the GFA. In this example, the group creation is static
in the sense that the numbers of groups and FAs do not
change and remain fix.
In Fig. 3, when the MT moves from 1
2
to 1
5
, the
location registration is performed between H and G1
2
.
G1
2
multicasts packets to 1
4
, 1
5
, and 1
6
. Thus,
when MT moves to 1
6
or 1
4
there is no need for the MT
location registration. Hence, this approach allows reducing
the mobility signaling delay compared to the HMIP and
DHMIP mobility approaches specifically for high-mobility
MTs. However, it is network resources consuming ap-
proach due to multicast protocol use. Consequently, it is
required for comparison purpose to evaluate the perfor-
mance not only in term of handoff signaling delay but also
in term of bandwidth use. This latest is the bandwidth used
for signaling transfer and packet delivery.
If we take the same MIP network architecture for the
three mobility management approaches, the bandwidth
used by MHMIP signaling is smaller than that of MIP or
1314 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
DHMIP approaches because the path reestablishment is
performed only between HA and GFAs. However, the
bandwidth used by an MT for packet delivery is high
because several connections are used for packets transfer to
the MT. It is clear that the total bandwidth used for
signaling and packet delivery in MHMIP approach is higher
than that used by the other approaches. Nevertheless, in
case of MTs with high mobility (high handoff requests), the
multicast resource in the GFA groups are reused by the MT
every handoff event that occurs during its call holding time.
Consequently, we expect that the MHMIP mean bandwidth
per call for MTs with high mobility is no greater than that of
the DHMIP and MIP mobility approaches. We also expect
that the MHMIP mean handoff delay (including signaling
and packet delivery delays) is smaller than that of the
DHMIP and MIP mobility approaches.
Hence, we propose to derive an analytic model that
allows computation of mean bandwidth and mean handoff
delay per call for MIP, DHMIP, and MHMIP mobility
approaches. These performance measurements are com-
puted according to the MTs mobility type (high or low) and
the call holding time duration. The model description and
the performance comparison of the three mobility ap-
proaches are discussed in the following sections.
3 ANALYTIC MODEL
This section describes the analytic model and the set of
established assumptions.
3.1 Assumptions
Generally, during each handoff, a path reestablishment is
required to maintain or to improve call quality. This
reestablishment uses signaling messages and involves a
change in the number of links of the mobile connection.
Note that the three mobility approaches described here
are based on a mobile connection path reestablishment
which leads to perform the following operations:
. CoA update with the HA,
. new path establishment from HA to FA for DHMIP
and MIP, and from HA to GFA for MHMIP,
. user data traffic transfer from the previous path to
the new one,
. previous path discard.
The DHMIP uses also path extension which requires
additional signaling messages to establish the path part
that extends the mobile connection from the previous FA to
the new one when the mobile move and becomes attached
to this latest.
Each connection is subjected to a certain number of
handoffs through its life duration (call holding time). This
latest is divided into i time intervals enough small to allow
the occurrence and the end of only one handoff during this
interval. In each time interval, we define
. c
o
as the probability that an FAs handoff (handoff
betweentwo FAs) occurs andends inthis interval and
. c
1
as the probability that the call ends in this interval.
The number of handoffs that could occur during a call
holding time depends on the MT dwelling time in a radio
cell and the traffic type: voice or data. Several voice traffic
researches have supposed that the dwelling time in a radio
cell is an exponential distribution [24], [25]. In fact, this
assumption depends on the shape of the radio cell and the
specific distributions of the mobiles speed and direction
which are difficult to characterize. In [26], [27], [28], [29],
[30], authors have demonstrated that the exponential
distribution for the dwelling time in radio cell is not
appropriated. They propose to replace it with complex
distributions such as Phase-Type, Lognormal, Hyperexpo-
nential, and HyperErlang requiring the identification of
several parameters related to the selected traffic model. In
order to simplify the computation of the mean bandwidth
and mean delay per call, we consider that the time between
the handoff events and the call duration is a geometric
distribution of mean 1c
o
1
and 1c
1
, respectively.
For data traffic, researches have addressed the problem
of the persistent congestion periods with non-negligible
packet losses [31], [32], [33], [34]. They show that these losses
do not allow the usage of Poisson model to model the TCP
traffic. In [33], [34], authors have demonstrated that the Self-
Similar processes are better models for TCP traffic mod-
elization than the exponential ones. However, in this study,
we are interested by the data session arrivals rather than the
data packet generation in the sessions. Hence, we propose
that the assumption made for the voice traffic remains valid
for the data traffic.
The proposed discrete time model is a generalization of
the one proposed in [35]. The novelty of this model consists
in the definition of generic analytical model that applies to
more than one handoff approach and that allows to
compute not only mean bandwidth due to handoff but also
mean handoff delay of the analyzed handoff approaches.
The temporal diagram given in Fig. 4 is used to compute
these means. First, we compute the bandwidth and the
delay within each interval and their means over the handoff
events. Then, we compute the bandwidth and the delay
sums over the total call holding time. Finally, we evaluate
their means over all the call durations. In order to
understand the modelization mechanism, we illustrate by
taking as an example the mean bandwidth computation. In
this figure, the holding time of ongoing call is divided into
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1315
1. The respective temporal means are obtained while multiplying by the
interval duration.
Fig. 3. Hierarchical handoff scheme.
time intervals small enough that we may assume that in
each time interval i. i 1, at most one handoff may occur.
In each interval, let
. i be the number of intervals for a call,
. 1
|
i
be the bandwidth used by a call during the time
interval i. i 1,
. 1
:
i
be the signaling bandwidth used by a call during
handoff that occurredinthe time interval i. i 1, and
. 1
i
be the total bandwidth used by a call during the
time interval i. i 1;
1
|
i
and 1
:
i
are random variables with values that depend on
the occurrence or not of a handoff during the interval i. i 1
and on the possible path reestablishment once the handoff
occurs. The variable 1
|
i
can take two values. When a handoff
occurs for a call in the interval i. i 1, 1
|
i
represents the sum
of the allocated bandwidth over the original path and the
one allocated over the links of the new established path.
Otherwise, it represents the bandwidth used on the link of
the ongoing connection. 1
i
represents the sum of the
bandwidth used by the ongoing call (1
|
i
) and the bandwidth
used for signaling (1
:
i
). Otherwise, it represents the allocated
bandwidth to the ongoing call (1
|
i
). Then, we obtain
1
i

1
|
i
1
:
i
. if a handoff occurs in i. i 1.
1
|
i
. otherwise.
&
1
The mean of 1
i
over the handoff events is given by
1

1
i

1
|
i

1
:
i

. 2
For fixed value of i, the total mean bandwidth 1i used
by an ongoing call during the i time intervals is
1i 1
|
i 1
:
i
X
i1
i0
1

1
|
i

X
i1
i0
1

1
:
i

. 3
As the call duration i is a random variable, the mean
bandwidth 1 is computed over all the call durations
i 1. . . . . 1. With our assumptions, the probability that a
call runs i periods is defined as 1i:
1i c
1
1 c
1

i1
i 1. 2. . . . 4
such as
i
X
1
i1
i1i
X
1
i1
ic
1
1 c
1

i1
c
1
X
1
i1
i1 c
1

i1
1c
1
represents the mean number of intervals during a call.
Then, we obtain
1 1
|
1
:

X
1
i1
1
|
i 1
:
i

1i

X
1
i1
1
|
i1i
X
1
i1
1
:
i1i.
5
Let 11
r
i
. r 2 f:. |g, a variable that designates the 11
|
i
or
11
:
i
entity. In the bandwidth computations given later,
11
r
i
could be variable or not during a call. If variable, then
we obtain
1
r

X
1
i1
1
r
i1i
X
1
i1
X
i1
i0
1

1
r
i

1i
c
1
X
1
i1
1 c
1

i1
X
i1
i1
1

1
r
i

;
6
otherwise, we have
1
r

X
1
i1
1
r
i1i
X
1
i1
X
i1
i0
1

1
r
i

1i

X
1
i1
1

1
r
i

X
i1
i0
1i 1

1
r
i

X
1
i1
i1i
c
1
1

1
r
i

X
1
i1
i1 c
1

i1

1

1
r
i

c
1
.
7
The same procedure applies for the mean handoff delay
computation by substituting to variable 1 the variable 1
which represents the delay. Note that 1 and 1 are random
variables due to handoff.
In the following sections, the mean bandwidth per call is
the network bandwidth needed to support a mobile
connection over its total duration, and it is given by the
sum of the bandwidth used on the paths links of the
ongoing connection and the signaling bandwidth due to
handoffs. Likewise, the mean handoff delay per call is the
network handoffs durations to support a mobile connec-
tion which is given by the sum of the duration of the
resource establishment on the paths links and the signaling
duration due to handoff. The paths links are the total
network links of all paths used by the ongoing connection
during its holding time. Let
. 1
11
be the allocated bandwidth on each link for
packet delivery of a call,
. 1
11
be the signaling bandwidth used per call for
each path extension,
. 1
11
be the signaling bandwidth used per call for
each path reestablishment,
. 1
11
be the duration per call to allocate bandwidth
1
11
on a link of a new extended path and/or a
reestablished path,
. 1
11
be the signaling duration per call for a path
extension, and
. 1
11
be the signaling duration per call for a path
reestablishment.
Note that 1
11
and 1
11
are, respectively, the bandwidth
and the duration for packet delivery on each link of a path
1316 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
Fig. 4. Discrete diagram of a call holding time.
for an ongoing call. The 1
11
can take different fixed values
according to the traffic type carried through the path links:
voice or data (e.g., 64 kbps). A link is the network
connection between two network entities such as an FA
and a router (e.g., in Fig. 3, there is three links between the
1
1
and HA). The signaling duration is the time taken for
the transmission and the execution of the different handoff
signaling messages. The parameters 1
11
and 1
11
are,
respectively, the bandwidth and the duration necessary to
set up the path extension between two FAs involved in a
handoff. The parameters 1
11
and 1
11
include the
bandwidth and the duration, respectively, for 1) CoA
registration, 2) setting up the new portion of connection
between HA and FA in case of DHMIP and MIP, and
between HA and GFA in case of MHMIP (e.g., path between
HA and 1
4
in Fig. 2), and 3) terminating the old portion of
a connection (e.g., path between HA and 1
3
in Fig. 2).
3.2 DHMIP Analytic Model
The DHMIP mobility approach combines the path rees-
tablishment and the connection extension protocols. The
path reestablishment protocol is invoked to set up a new
FAs hierarchy. This protocol allows a path establishment
between the HA and a new FA in the new hierarchy. In
this latest, the path extension is used to maintain the
mobile connection when mobile moves through the FAs
belonging to this hierarchy. The path reestablishment may
occur after each new FAs hierarchy setup. Events that may
occur at each time i 1. 2. . . . are 1) path reestablishment,
2) path extension, and 3) call termination. Let
. j be the probability that a newFAhierarchy is set and
consequently a path reestablishment is performed,
. 1 be the number of links between the FA to which
the MT is attached and the remote end point with
which the MT communicates,
. 1
j
be the number of links between the HA and the
initial FA through which a new hierarchy is set (e.g.,
1
1
and 1
4
in Fig. 2), and
. H be the number of links of the path extension
(e.g., in Fig. 2, this number is equal to 1 when MT
moves from 1
1
to 1
2
and becomes connected
to 1
2
).
1, 1
j
, and H are random variables with general distribu-
tions and with mean 1, 1
j
, and H, respectively.
The mean bandwidth per call is
1
j

1
c
1
1
11

c
o
1 j1 c
1
H
c
1
1 1 jc
o
1 c
1

1
11

c
o
c
1
1
11
j1
11
.
8
while the mean handoff delay per call is
1
j

c
o
c
1
1
11
1 jH j1
j

c
o
c
1
1
11
j1
11
.
9
In (8), the first term (
1
c
1
1
11
) represents the bandwidth
used on the original path and the paths resulting from the
reestablishment. The second term (
co1j1c1 H
c
1
11jc
o
1c
1

1
11
)
represents the additional bandwidth due to the path
extensions. The last term (
c
o
c
1
1
11
j1
11
) represents the
signaling bandwidth due to the extensions and path
reestablishments.
In (9), the term c
o
c
1
represents the mean number of
handoffs of a call. The second term 1
11
1 jH j1
j

1
11
j1
11
represents the handoff delay which is the sum
of the resource reservation delay on the links of the extended
and the reestablished paths (1
11
1 jH j1
j
), and the
signaling delay due to the path extensions and the path
reestablishments (1
11
j1
11
).
The reader is referred to Appendix A.1 and B.1 for a
detailed demonstration of these formulas.
3.3 MIP Analytic Model
The MIP mobility approach is based only on the path
reestablishment protocol. This latest allows maintaining
the call connectivity when the MT moves between FAs. In
this case, events that may occur at each time i 1. 2. . . .
are 1) path reestablishment and 2) call termination. Let
. c
o
be the probability that there is an inter-FAs
handoff and thus a partial reestablishment,
. 1 be the number of links between the FA to which
the MT is attached and the remote end point with
which the MT is communicating, and
. 1
i
be the number of links between the HA and the
new FA to which the MT moved (e.g., the number of
links between the HA and the 1
3
following the
handoff from 1
1
to 1
3
in Fig. 1).
1 and 1
i
are random variables with general distributions
and with mean 1 and 1
i
, respectively.
The mean bandwidth per call is
1
i

1
c
1
11
11

c
o
c
1
1
11

. 10
In (10), the first term
1
c
1
11
11
is the bandwidth of the
original connection and the reestablished paths. The second
term
co
c
1
1
11
is the signaling bandwidth due to the path
reestablishments.
The mean handoff duration per call is
1
i

c
o
c
1
1
i
1
11
1
11
. 11
In(11), theterm
co
c
1
represents the meannumber of handoffs
for a call. The term 1
i
1
11
1
11
represents the handoff
delay which is the sumof the delay for resource allocation on
thereestablishedpath(1
i
1
11
) andthe signalingdelay(1
11
).
Details on these computations are given in Appendix A.2
and B.2.
3.4 MHMIP Analytic Model
The MHMIP mobility approach is based on the path
reestablishment and the multicast protocols. When the MT
moves within a GFA group, the mobile connection is
maintained using the multicast protocol. When the MT
moves outside this hierarchy, a combination of the path
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1317
reestablishment and the multicast protocols allows main-
taining the calls connection. Events that may occur at
each time i 1. 2. . . . are 1) path reestablishment and
2) call termination.
We define c
0
o
as the probability that there is an inter-
GFAs handoffs and thus path reestablishments such as
c
0
o
0c
o
with 0 0 1. 0 is the fraction of inter-GFAs
MHMIP handoffs on the whole possible handoffs c
o
(intra-
and inter-GFAs).
The inter-GFAs handoff arrivals are modeled using a
Bernoulli process. For each mobile connection, we define
. 1
/
as the number of links between the GFA to which
the mobile is currently attached and the remote end
point with which the MT is communicating,
. 1
/j
as the number of links between the HA and the
GFA to which the mobile is currently belonging, and
. 1
/i
as the total number of links inthe GFAhierarchies.
1
/
, 1
/j
, and 1
/:
are random variables with general
distributions and with means 1
/
, 1
/j
, and 1
/:
, respectively.
The mean bandwidth per call is
1
/

1
c
1
1
/
1
11
1
/i
1
11

c
0
o
c
1
1
11
. 12
In (12), the first term
1
c
1
1
/
1
11
is the bandwidth used on
the original path and the reestablished paths. The second
term 1
/i
1
11
is associated to the multicast resources used by
the call in the GFA hierarchies. The last term
c
0
o
c
1
1
11
is the
signaling bandwidth due to the path reestablishment
following the GFA handoffs.
The mean call duration per call is
1
/

c
0
o
c
1
1
/j
1
11
1
11
. 13
In (13), the term
c
0
o
c1
is the mean number of handoffs of a
call. The second term 1
/j
1
11
1
11
is the handoff delay
which is the sum of the delay of resource allocated on the
reestablished path (1
/j
1
11
) and the signaling delay (1
11
).
The details on these computations are given in
Appendix A.3 and B.3.
4 RESULTS ANALYSIS
In this section, we compare the performance in terms of
mean bandwidth and mean handoff delay per call of the
three mobility management approaches MHMIP, DHMIP,
and MIP.
4.1 Numerical Data
The mean call holding time is a random value chosen
between 60 and 120 seconds for voice traffic and between
900 and 1,200 seconds for data traffic. For simplification
purpose of the mean number of links computation (1
i
, 1
j
,
1, H, 1
/
, 1
/j
, and 1
/i
), a symmetric hierarchical IP network
architecture is considered (Fig. 5). Symmetric architecture
means that the number of links between the HA and each
FA is the same (e.g., there is five links between the HA and
each 1
i
. fi 1. . . . . 32g in Fig. 5). The example given in
Fig. 5 shows an architecture with 1
j
1
i
7, 1
/j
3, and
1
/i
240.
For comparison purpose, we take the number of links
between the HA and the end point the same for the three
mobility management approaches. For a fixed remote end
point, the number of links between the HA and this end
point do not change for an ongoing call of an MT. Then, we
consider that the end point is directly connected to HA
(e.g., 1
/
1
/j
3 and 1 1
j
1
i
7 for the example
given in Fig. 5).
Two types of configurations are considered for the
network given in Fig. 5:
. Configuration 1: the average number of links are
1
/
1
/j
3 and 1
i
1
j
1 7. These values
result in the number of link where the resources
were allocated 1
/i
240.
. Configuration 2: the average number of links are
1
/
1
/j
1 and 1
j
1
i
1 7. From these
values, we obtain 1
/i
252.
For each configuration, two cases are analyzed: realistic and
critical. In the realistic case, the inter-GFAs handoffs may
occur less frequently than the intra-GFAs handoffs
(c
0
o
0.1 c
o
). In the critical case, the intra- and the inter-
GFAs handoffs may occur with the same probability
(c
0
o
c
o
, where c
o
and c
0
o
are variables).
For both cases, the path extension for the DHMIP
mobility management approach should occur after each
handoff and the path reestablishment should occur after
each two consecutive handoffs (j c
o
2). For j c
o
2, the
mean bandwidth and mean delay is higher than that get
with j c
o
2 (see Section 4.2).
We suppose that the MT handoff to a new FA involves a
path extension of mean length H 1. For length greater
than this value, the mean bandwidth and the mean handoff
delay are high.
We rewrite (8), (12), (10), (9), (13), and (11) to obtain the
ratios 1
,
11
1
,
1
11
and 1
,
11
1
,
1
11
, where , j. i. /.
1318 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
Fig. 5. Symmetric hierarchical IP network architecture.
Thus, these equations contain the following ratios: 1
11
1
11
,
1
11
1
11
, 1
11
1
11
, and 1
11
1
11
. We expect that the
signaling bandwidth for the path extension in DHMIP is
smaller than that of the path reestablishment (1
11
1
11
and 1
11
1
11
). Then, 1
11
1
11
< 1 and 1
11
1
11
< 1.
As an example, we take 1
11
1
11
1
11
1
11
0.2.
Moreover, we expect that 1
11
1
11
and 1
11
1
11
.
Indeed, the 1
11
represents the sum of the allocated
bandwidth on each link of the path over which the handoff
signaling is transferred (the path between HA and FA for
DHMIP and MIP, and between HA and GFA for MHMIP).
This bandwidth could be greater than 1
11
which is the
bandwidth allocated on a path link for packet transfer.
According to the 1
11
and the signaling bandwidth values,
we could have 0 < 1
11
1
11
< 1. In the same way, 1
11
represents the sum of the delays for bandwidth allocation on
each link of the path carrying the signaling traffic and the
delay for signaling messages processing. It is greater than
1
11
, which represents the delay for 1
11
allocation. Thus, we
take 0 < 1
11
1
11
< 1. In this analysis, we showan example
of the results for 1
11
1
11
1
11
1
11
0.5 and 0.8.
4.2 Numerical Results
We propose to compare the performance of the MHMIP
handoff approach with those obtained with DHMIP and
MIP approaches in terms of mean bandwidth and mean
handoff delay per call. For summarization purpose, we
compute the ratios 1
j
11
1
/
11
, 1
i
11
1
/
11
, 1
j
11
1
/
11
, and
1
i
11
1
/
11
. These ratios allow a simple and direct reading of
the different performance between the tree mobility
management approaches.
Figs. 6 and 7 give an example of mean bandwidth
variation per call 1
j
11
and 1
/
11
for the DHMIP and MHMIP
handoff approaches.
Fig. 6 illustrates the mean bandwidths per call for
MHMIP and DHMIP mobility management approaches. It
shows that the MHMIP mean bandwidth per call is smaller
than that obtained with the DHMIP approach. This mean
bandwidth represents a performance measurement that an
IP network operator can use to determine the needed
resources to be deployed in the network to service a certain
number of MTs. The MHMIP mobility management
approach is the method that allows cost reduction in terms
of resources usage compared to the DHMIP approach.
Fig. 7 illustrates the 1
j
1
11
ratio variation for different
values of the probability j. We note that lower is j higher is
the mean bandwidth per call. Moreover, we note a different
behavior of this bandwidth between the intervals c
o
0.3
and 0.3 c
o
1. For 0.3 c
o
1, the mean bandwidth
value decreases while it increases in the interval c
o
0.2 for
different values of j (j c
o
6. c
o
4. c
o
2) and still increasing
in the interval 0.2 c
o
0.3 for j c
o
6. This is in fact due
to the low probability of path reestablishment j and the
frequent use of path extension in the interval c
o
0.3.
Hence, less frequent path reestablishment usage for DHMIP
mobility management approach involves a high mean
bandwidth per call consumption.
4.2.1 Mean Bandwidth
Figs. 8 and 9 show examples of the mean bandwidth ratio
variation 1
j
11
1
/
11
and 1
i
11
1
/
11
for the realistic and the
critical cases, respectively.
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1319
Fig. 6. Mean bandwidth per call 1
j
11
and 1
/
11
for voice traffic with 1c
1

60 seconds, 1
11
1
11
0.5.
Fig. 7. Mean bandwidth per call variation 1
j
11
for voice traffic with 1c
1

60 seconds, 1
11
1
11
0.5.
Fig. 8. Mean bandwidth ratio 1
j
11
1
/
11
and 1
i
11
1
/
11
for voice traffic
with j c
o
2. 1
11
1
11
0.8.
Note that the ratio 1
j
11
1
/
11
is much higher than the
ratio 1
i
11
1
/
11
for different call holding time duration
(1c
1
60. 90. 120 seconds) specifically for small probabil-
ity c
o
. This means that the combination of the path
extension and the path reestablishment for handoff
management involves higher mean bandwidth per call
than that used by the approaches based only on the path
reestablishment (such as MHMIP and MIP). This behavior
is noticed in all the analyzed cases. The main obtained
results are summarized in Table 1 that give the ratio values
of the mean bandwidth per call for the Realistic Case (RC)
and the Critical Case (CC) for both Types of Configuration
(ToC). These values represent the arithmetic average of the
ratios 1
j
11
1
/
11
and 1
i
11
1
/
11
over the set of c
o
values
(0 c
o
1). These results show that the MHMIP mean
bandwidth is smaller than those of the DHMIP and MIP
approaches. This bandwidth difference is higher in the
configuration 2 than in the configuration 1 because the
MHMIP reestablishment is performed over small number
of links in the configuration 2, yielding to a smaller mean
bandwidth per call than that computed with the config-
uration 1. However, this mean bandwidth difference is
small in the critical case than in the realistic case because
the partial reestablishment is more frequently used due to
the high probability c
0
o
of inter-GFAs handoffs.
4.2.2 Delay Comparison
The ratio values of the mean delay per call 1
j
11
1
/
11
and
1
i
11
1
/
11
are summarized in Table 2 for the Realistic Case
(RC) and the Critical Case (CC). These values represent the
arithmetic average of the ratios 1
j
11
1
/
11
and 1
i
11
1
/
11
over the set of c
o
values. In the realistic case, the MHMIP
mean delay is smaller than those of the DHMIP and MIP
approaches. The mean delay differences are 8 and 21
compared to DHMIP and MIP, respectively. They become
high, if we consider the configuration 2 where this difference
reaches 40. This result was expected because in the MHMIP
approach, the path reestablishment is performed through a
shorter path than that of the DHMIP and MIP approaches.
The mean delay per call of the MHMIP approach in the
critical case is smaller than that of the MIP and DHMIP
approaches unless for configuration 1 where this delay is
greater than that of the DHMIP approach, because the
path reestablishment is not only more frequent (c
0
o
c
o
)
but also the number of links involved in the path
reestablishment is greater than that of configuration 2
(1
/
3 compared to 1
/
1).
4.3 Recommendation
Hence, we can derive the following recommendations that
indicate when one would use the MHMIP, the DHMIP, or
the MIP mobility management approaches:
. The MHMIP versus DHMIP usage:
1. If the inter-GFAs handoffs are not frequent then
we suggest to use the MHMIP approach that
provides a best mean handoff delay and mean
bandwidth per call for voice and data traffic.
2. If the inter-GFAs handoffs occur frequently
(such as c
0
o
c
o
), then we have two cases. If the
mean handoff delay per call is more important
than the mean bandwidth per call and if the
number of links involved in MHMIP path
reestablishment is high, then we suggest using
the DHMIP handoff approach; otherwise, we
suggest using the MHMIP handoff approach.
1320 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
Fig. 9. Mean bandwidth ratio 1
j
11
1
/
11
and 1
i
11
1
/
11
for data traffic with
j c
o
2. 1
11
1
11
0.8.
TABLE 1
Mean Bandwidth
TABLE 2
Mean Delay
. The MHMIP versus the MIP usage: We suggest
using the MHMIP handoff approach that gives a best
mean handoff delay per call and a mean bandwidth
per call for voice and data traffic, and for both
frequent and nonfrequent GFAs handoffs.
5 CONCLUSION
In this paper, we have proposed an analytical model
which evaluates the mean handoff delay per call and the
mean bandwidth per call of three mobility management
approaches: MIP, DHMIP, and MHMIP. Numerical results
show that the MHMIP mobility approach compares very
favorably with the previously considered mobility ap-
proaches. More specifically, our analysis gives in almost all
cases a lower mean handoff delay per call and a mean
bandwidth per call than those offered by the DHMIP and
MIP approaches. It also shows the robustness of the
MHMIP approach in the sense that for critical scenario
corresponding to the extreme situation where all handoff
events are localized at the multicast group borders, this
approach essentially yields to 1) a lower mean bandwidth
per call than the DHMIP and MIP approaches; 2) a lower
mean handoff delay per call than that offered by the MIP
approach; 3) a lower mean handoff delay than that offered
by the DHMIP except in case of frequent inter-GFAs
handoffs with a network configuration having a high
number of links involved in MHMIP path reestablishment
such as the configuration 2. Since we expect a diversity of
multimedia applications for future IP mobile networks, we
recommend using the MHMIP approach in networks parts
carrying delay sensitive and/or low mean bandwidth
consumption type of applications and this according to
the mobility type.
APPENDIX A
EVALUATION OF THE MEAN BANDWIDTH PER CALL
By using the discrete time model given in Fig. 4, we
estimate the mean bandwidth per call for the DHMIP, MIP,
and MHMIP mobility management approaches.
A.1 DHMIP Approach
In this case, we take into account the following events: the
inter-FAs handoffs (path extension) that occur with prob-
ability c
o
and the path reestablishment which is executed
with probability j.
For each time interval i. i 1, we define two random
variables corresponding to the occurrence or not of a
handoff event and the path reestablishment.
1
i

1. a handoff occurs.
0. otherwise.
&
14
J
i

1. a path reestablishment is executed.
0. otherwise.
&
15
We also suppose that 11
i
c
o
and 1J
i
j.
The mean bandwidth used by a call in a time interval
i. i 1 is given by
1
|
i
1
|
i1
1 1
i
1
i
1 J
i

11
i
J
i
1
i
i 1 J
i
H
i
i
1
|
i1
1 1
i
J
i

11
i
J
i
1
i
i 1 J
i
H
i
i .
16
We define new random variables :
,
1
,
. J
,
, where :
,
represents the pair of variables 1 and J at the period , and
the vector s
j
:
1
. :
2
. . . :
,
.
At each interval, 1
i
and H
i
represent, respectively, the
number of links of the original path or a path resulting
from a path reestablishment and the number of links of
the extended path. In general, these values are depen-
dent in complex way of events that occurred before the
instant i. They are generally random variables which
should be designated by 1
i
s
i
and H
i
s
i
. Then, we can
rewrite (16) more explicitly such as
1
|
i
s
i
1
|
i1
s
i1
1 1
i
J
i

11
i
J
i
1
i
s
i
1 J
i
H
i
s
i
.
17
As the events are independent in each interval, we can write
for s
i
js
i
js
i1
j1
i
. J
i
.
where j1
i
. J
i
is the probability to have a given config-
uration of the random variables 1
i
and J
i
. Note that
j0. 1 0 because we could have a path reestablishment
only after a handoff. Let define the two auxiliary entities:
o1
i
. J
i
1 1
i
J
i
. 18
/s
i
11
i
J
i
1
i
s
i
1 J
i
H
i
s
i

. 19
The mean bandwidth over the handoff events is
given by
1

1
|
i

X
s
i
js
i
7
i

X
s
i
js
i1
j1
i
. J
i
7
i

X
s
i1
js
i1

X
1i.Ji
j1
i
. J
i
7
i
.
20
where 7
i
1
i1
s
i1
o1
i
. J
i
/s
i
. As 1
i
and H
i
are
dependent on s
i
in a complex way and for computation
simplification purpose, a set of assumptions are made. We
suppose these variables constant from the fact that in
general we can expect that the connection paths and the
path extensions do not vary a lot from one handoff to
another. From this assumption, we can continue our
computations such as (20) becomes
1

1
|
i

X
s
i1
js
i1
1
i1
s
i1

X
1
i
.J
i
j1
i
. J
i
o1
i
. J
i

X
s
i1
js
i1

X
1i.Ji
j1
i
. J
i
/1
i
. J
i

1
|
i1

X
1
i
.J
i
j1
i
. J
i
o1
i
. J
i

X
1
i
.J
i
j1
i
. J
i
/1
i
. J
i
.
21
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1321
with
P
s
i1 js
i1
1. We can compute the two terms:
X
1i.Ji
j1
i
. J
i
o1
i
. J
i
1 c
o
o0. 0 c
o
jo1. 1
c
o
1 jo1. 0.
22
X
1i.Ji
j1
i
. J
i
/1
i
. J
i
1 c
o
/0. 0 c
o
j/1. 1
c
o
1 j/1. 0.
23
with o0. 0 1. o1. 1 0, o1. 0 1, /1. 1 1, /1. 0
H, and /0. 0 0. By replacing each of these terms in (21),
we obtain
1

1
|
i

1 c
o
j1

1
|
i1

c
o
1 jH j11
11
. 24
If we suppose that 11
|
0
11
11
, we can solve the
recurrence (24) and we obtain
1

1
|
i

1 c
o
j
i
11
11
1 1 c
o
j 1 c
o
j
i1
`
1 c
o
j
i
11
11

1 1 c
o
j
i
c
o
j
`.
25
where ` c
o
1 jH c
o
j11
11
.
The mean signaling bandwidth due to handoff in the
interval i. i 1 is given by
1
:
i
1
i
1
11
J
i
1
11
. i ! 1. 26
with the mean of 1
:
i
over the handoff events is
11
:
i
c
o
1
11
j1
11
. i ! 1. 27
As 11
:
i
is not a variable, then we use (7) and (27) to
compute 1
:
.
1
:

1
:
i

c
1
c
o
1
11
j1
11

1
c
1
. 28
We use (25) such as
1
|
i
X
i1
i0
`
c
o
j
11
11

`
c
o
j
!
X
i1
i0

`i
c
o
j
11
11

`
c
o
j
!
1
i
c
o
j

.
29
where 1 c
o
j.
Using (6) and (29), we find
1
|
c
1
X
1
i1
1 c
1

i1
X
i1
i1
1

1
|
i

`
c
o
j
X
1
i1
i1i 1
1
c
o
j
X
1
i1
1
i
1i.
30
where 1 11
11

`
c
o
j
. We evaluate the following
equation:
X
1
i1

i
1i c
1
X
1
i1

i
1 c
1

i1

c
1
1 c
1
X
1
i1
1 c
1

i
c
1

1 1 c
1

!
.
31
By replacing in (30), we obtain
1
|

`
c
o
jc
1
1
1
c
o
j
1
c
1

1 1 c
1

`
c
o
jc
1
1
1
c
o
j
c
o
j
1 1 c
1

`
c
o
jc
1
1
1
1 1 c
o
j1 c
1

!
.
32
By replacing ` with its value and by reorganizing the
terms we finally find
1
|

1
c
1
1
11
1
11
1 jH
1
c
1
c
o
1 c
1

1 1 c
o
j1 c
1

!
.
33
Using (28) and (33), we compute 1
j
1
|
1
:
1
j

1
c
1
1
11

c
o
1 j1 c
1
H
c
1
1 1 c
o
j1 c
1

1
11

c
o
c
1
1
11
j1
11
.
34
A.2 MIP Approach
An inter-FAs handoff occurs with probability c
o
. For each
time interval i. i 1, we define the random variable
1
i

1. a handoff occurs.
0. otherwise.
&
35
which corresponds to the occurrence of a handoff followed
by a path reestablishment of the MTs ongoing connection.
Note that 11
i
c
o
.
The mean signaling bandwidth due to mobility in the
interval i. i 1 is given by
1
:
i
1
i
1
11
. i ! 1. 36
The mean bandwidth corresponding to the handoff events
is given by
1

1
:
i

c
o
1
11
. i ! 1. 37
The bandwidth used in the time interval i. i 1 is
given by
1
|
i
1 1
i
1
|
i1
1
i
i1
i
1
11
. i ! 1. 38
Note that 1
|
i
depends on all the random variables 1
,
. ,
1. . . . . i 1 and on the variable 1
i
of the current interval.
The length 1
i
i depends in complex manner on the events
that occurred before i. It is represented by 1
i
s
i
with
s
i
1
1
. 1
2
. . . . . 1
i
. We can then rewrite (38) as
1
|
i
s
i
1 1
i
1
|
i1
s
i1
1
i
s
i
1
i
1
11
. i ! 1. 39
1322 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
Let js
i
js
i1
j1
i
, where j1
i
is the probability of the
occurrence or not of a handoff such as j1 c
o
and
j0 1 c
o
, and let the auxiliary entities o1
i
1 1
i

and /s
i
1
11
1
i
1
i
s
i
. The mean value on the handoff
events is given by
1

1
|
i

X
s
i
js
i
1
X
s
i
js
i1
j1
i
1

X
s
i1
js
i1

X
1
i
j1
i
1.
40
where 1 1
i1
s
i1
o1
i
/s
i
. As 1
i
depends in com-
plex manner on s
i
, we assume that 1
i
is constant. This
assumption is valid since the connection length after a path
reestablishment doesnt change a lot from one handoff to
another. Consequently, we can write
1

1
|
i

X
s
i1
js
i1
1
i1
s
i1

X
1
i
j1
i
o1
i

X
s
i1
js
i1

X
1
i
j1
i
/1
i

1
|
i1

X
1
i
j1
i
o1
i

X
1
i
j1
i
/1
i
.
41
with
P
s
i1 js
i1
1. We can compute the two terms
X
1
i
j1
i
o1
i
1 c
o
o0 c
o
o1. 42
X
1i
j1
i
/1
i
1 c
o
/0 c
o
/1. 43
with o0 1. o1 0. /0 0, and /1 1. By replacing
these terms in (41), we find
1

1
|
i

1 c
o
1

1
|
i1

c
o
11
11
. 44
Let 11
|
0
11
11
, we can solve the recurrence and we
obtain
1

1
|
i

1 c
o

i
11
11

1 1 c
o

i
c
o
A 11
11
. 45
with A c
o
11
11
.
As 11
:
i
given in (37) is not variable, then we compute
1
:
using (7):
1
:
c
o
1
11
1
c
1
. 46
Since 11
:
i
given in (45) is not variable, we use (7) and (45)
to compute 1
|
.
1
|
11
1
c
1
47
We use (46) and (47) to compute the mean bandwidth per
call such as
1
i
1
|
1
:

1
c
1
11
11
c
o
1
11
. 48
A.3 MHMIP Approach
In this case, two types of handoff can occur in each time
interval: inter and intra-GFAs handoffs. From the fact that
intra-GFAs handoff doesnt generate signaling traffic. This
type of events is not considered in our computations.
In each time interval i. i 1, we define a random
variable corresponding to the inter-GFAs handoff occur-
rence and consequently to a path reestablishment.
1
i

1. an inter-GFAs handoff occurs.
0. otherwise.
&
49
with a mean value 11
i
c
0
o
.
The computation of the mean bandwidth per call follows
the same steps than those explain for the MIP handoff
approach by replacing c
o
by c
0
o
and 1
i
by 1
/
i
.
The mean bandwidth per call for the MHMIP handoff
approach is given by
1
/

1
c
1
1
/
1
11
c
o
1
11
. 50
To this bandwidth, we add a fixed bandwidth due to
resources allocation in the GFAs hierarchies 1
/:
given by
1
/:
1
/:
1
11
. 51
1
/:
is the number of links in the GFAs hierarchies. Then,
the mean bandwidth per call is given by
1
/

1
c
1
1
/
1
11
1
/:
1
11
c
o
1
c
1
1
11
. 52
APPENDIX B
EVALUATION OF MEAN HANDOFF DELAY PER CALL
As in Appendix A, we consider the temporal diagram given
in Fig. 4. The handoff delay is defined as the sum of the
delay due to bandwidth allocation 1
11
on the new path
links and the signaling delay. In each interval i. i 1, let
. 1
|
i
be the delay to allocate the bandwidth 1
11
on a
link of the new path resulting from a handoff event
during a call,
. 1
:
i
be the signaling delay due to a handoff that
occurred in the time interval i. i 1, and
. 1
i
be the total delay for a call in the time
interval i. i 1.
1
|
i
and 1
:
i
are the random variables which depend on the
occurrence of handoff events in the interval i. i 1. 1
i
represents the sum of the delay due to the bandwidth
allocation on the new path following a handoff event and
the signaling delay.
1
i
1
|
i
1
:
i
. 53
B.1 DHMIP
We reuse the same procedure than that used for the
computation of the DHMIP mean bandwidth per call
(Appendix A.1) by replacing 1 by 1.
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1323
The signaling delay due to handoff event in the time
interval i. i 1 is given by
1
:
i
1
i
1
11
J
i
1
11
. i ! 1. 54
The mean 1
:
i
, on the handoff events is given by
1

1
:
i

c
o
1
11
j1
11
. i ! 1. 55
The delay for bandwidth 1
|
i
allocation in the time
interval i. i 1 is given by
1
|
i
H
i
1
11
1
i
1 J
i
1
j
i
1
11
1
i
J
i
. 56
Note that 1
|
i
depends on the random variables 1
i
and J
i
in
the current interval. If we assume that 1
j
i
and H
i
do not
change a lot according to i and in general they remain
constant, we obtain
1

1
|
i

c
o
1 jH j1
j
1
11
. 57
The handoff delay 1i during the i periods of a call is
1
j
i 1
|
i 1
:
i
X
i1
i0
1 1
|
i

X
i1
i0
1 1
:
i

c
o
i 1 jH j1
j
1
11
1
11
j1
11

.
58
The mean handoff delay per call is given by
1
X
1
i1
1
:
i 1
|
i1i
c
o
j
1 jH1
11

c
o
j
j1
j
1
11
1
11
j1
11

.
59
B.2 MIP
To compute the handoff delay, we reuse the same
procedure described in Appendix A.2 for the computation
of the MIP mean bandwidth per call. The signaling delay
due to handoff event in the time interval i. i 1 is
1
:
i
1
i
1
11
. 60
The mean handoff delay of 1
:
i
over the handoff events is
given by
1

1
:
i

c
o
1
11
. 61
The delay due to the bandwidth 1
|
i
allocation is given by
1
|
i
1
j
i
1
11
1
i
. i ! 1. 62
We assume that 1
j
i
is constant, we obtain
1
c
o

1
|
i

c
o
1
j
1
11
. 63
The handoff delay 1
:
i during the i periods of a call is
given by
1i 1
|
i 1
:
i
X
i1
i0
1

1
|
i

X
i1
i0
1 1
:
i

c
o
1
j
1
11
i c
o
1
11
i.
64
The mean handoff delay per call is given by
1
X
1
i1
1
:
i 1
|
i1i
c
o
c
1
1
j
1
11
1
11
. 65
B.3 MHMIP
The handoff delay computation procedure is similar to that
of MIP approach. By replacing 1
j
by 1
/j
, the mean handoff
delay per call becomes
1
j

X
1
i1
1
:
i 1
|
i1i

c
o
c
1
1
/j
1
11
1
11
.
66
REFERENCES
[1] C.E. Perkins, IP Mobility Support for IPv4, IETF RFC 3344, Aug.
2002.
[2] D. Johnson, C. Perkins, and J. Arkko, Mobility Support in IPv6,
IETF RFC 3775, June 2004.
[3] R. Caceres and V.N. Padmanabhan, Fast and Scalable Handoffs
for Wireless Internetworks, Proc. ACM MobiCom, pp. 56-66, 1996.
[4] C. Castelluccia, Extending Mobile IP with Adaptative Indivi-
dual Paging: A Performance Analysis, Proc. Fifth IEEE Symp.
Computers and Comm., pp. 113-118, July 2000.
[5] H. Soliman, C. Castelluccia, K. El-Malki, and L. Bellier, Hier-
archical Mobile IPv6 Mobility Management HMIPv6, IETF RFC
4140, Aug. 2005.
[6] E. Fogelstroem, A. Jonsson, and C. Perkins, Mobile IPv4 Regional
Registration, IETF RFC 4857, June 2007.
[7] H. Omar, T. Saadawi, and M. Lee, Supporting Reduced Location
Management Overheadand Fault Tolerance in Mobile IP Systems,
Proc. IEEE Symp. Computers and Comm., pp. 347-353, 1999.
[8] S. Pack, T. You, and Y. Choi, Performance Analysis of Robust
Hierarchical Mobile IPv6 for Fault-Tolerance Mobile Services,
IEICE Trans. Comm., vol. E87-B, no. 5, pp. 1158-1165, May 2004.
[9] J. Xie and I.F. Akyildiz, A Novel Distributed Dynamic Location
Management Scheme for Minimizing Signaling Costs in Mobile
IP, IEEE Trans. Mobile Computing, vol. 1, no. 3, pp. 163-175, July
2002.
[10] M. Song, J. Huang, R. Feng, and J. Song, A Distributed Dynamic
Mobility Management Strategy for Mobile IP Networks, Proc.
Sixth Intl Conf. ITS Telecomm., 2006.
[11] L. Yu, W. Yu-Mei, and Z. Hui-Min, Modeling and Analyzing the
Cost of Hierarchical Mobile IP, Proc. Intl Conf. Wireless Comm.,
Networking and Mobile Computing, vol. 2, pp. 1102-1105, Sept. 2005.
[12] 3GPP-TR-23.923, Combined GSM and Mobile IP Mobility
Handling in UMTS IP CN, technical report, May 2000.
[13] 3GPP-TR-25.913, Requirements for Evolved UTRA (E-UTRA)
and Evolved UTRAN (E-UTRAN), technical report, Mar. 2006.
[14] 3GPP-TR-23.882, 3GPP System Architecture Evolution: Report on
Technical Options and Conclusions (Release 8), technical report,
Sept. 2008.
[15] 3GPP-TR-33.922, Security Aspects for Inter-Access Mobility
between Non 3GPP and 3GPP Access Network (Release 8),
technical report, Sept. 2008.
[16] A. Acampora and M. Naghshineh, An Architecture and
Methodology for Mobile-Executed Handoff in Cellular ATM
Networks, IEEE J. Selected Areas in Comm., vol. 12, no. 8,
pp. 1365-1375, Oct. 1994.
[17] S. Seshan, Low-Latency Handoffs for Cellular Data Networks,
PhD dissertation, Univ. of California, Mar. 1996.
[18] R. Ghai and S. Singh, An Architecture and Communication
Protocol for Picocellular Networks, IEEE Personal Comm., vol. 1,
no. 3, pp. 36-46, 1994.
[19] M. Stemm and R. Katz, Vertical Handoffs in Wireless Overlay
Networks, ACM Mobile Networking (MONET), special issue on
mobile networking in the Internet, vol. 3, no. 4, pp. 335-350, 1998.
[20] J. Wu and G. Maguire, Agent-Based Seamless IP Multicast
Receiver Handover, Proc. Conf. Personal Wireless Comm. (PWC
00), pp. 213-225, Sept. 2000.
1324 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 8, NO. 10, OCTOBER 2009
[21] A. Mihailovic, M. Shabeer, and A.H. Aghvami, Multicast for
Mobility Protocol (MMP) for Emerging Internet Networks, Proc.
11th IEEE Intl Symp. Personal, Indoor, and Mobile Radio Comm.
(PIMRC 00), Sept. 2000.
[22] A. Helmy, M. Jaseemuddin, and G. Bhaskara, Multicast-Based
Mobility: A Novel Architecture for Effecient Micromobility, IEEE
J. Selected Areas in Comm. (JSAC), special issue on all IP wireless
networks, vol. 22, no. 4, pp. 677-690, May 2004.
[23] A. Festag, H. Karl, and A. Wolisz, Investigation of Multicast-
Based Mobility Support in All IP Cellular Networks, Wireless
Comm. and Mobile Computing, vol. 7, no. 3, pp. 319-339, Mar. 2007.
[24] B. Jabbari, Teletraffic Aspects of Evolving and Next-Generation
Wireless Communications Networks, IEEE Personal Comm.,
vol. 3, no. 6, pp. 4-9, Dec. 1996.
[25] S. Su, J. Chen, and J. Huang, Performance Analysis of Soft
Handoff in CDMA Cellular Networks, IEEE J. Selected Areas in
Comm., vol. 14, no. 9, pp. 1762-1769, Dec. 1996.
[26] A.S. Alfa and W. Li, A Homogeneous PCS Network with Markov
Call Arrival Process and Phase Type Cell Residence Time,
Wireless Networks, vol. 8, no. 6, pp. 597-605, Nov. 2004.
[27] Y. Fang and I. Chlamtac, Teletraffic Analysis and Mobility
Modeling of PCS Networks, IEEE Trans. Comm., vol. 47, no. 7,
pp. 1062-1072, July 1999.
[28] R. Guerin, Channel Occupancy Time Distribution in a Cellular
Radio System, IEEE Trans. Vehicular Technology, vol. VT-36, no. 3,
pp. 89-99, Aug. 1987.
[29] C. Jedrzycki and V. Leung, Probability Distributions of Channel
Holding Time in Cellular Technology Systems, Proc. IEEE
Vehicular Technology Conf., pp. 247-251, 1996.
[30] P. Orlik and S. Rappaport, A Model for Teletraffic Performance
and Channel Holding Time Characterization in Wireless Cellular
Communication with General Session and Dwell Time Distribu-
tions, IEEE J. Selected Areas in Comm., vol. 16, no. 5, pp. 788-803,
June 1998.
[31] P. Danzig, S. Jamin, R. Caceres, D. Mitzel, and D. Estrin, An
Empirical Workload Model for Driving Wide Area TCP/IP
Network Simulations, Interworking: Research and Experience,
vol. 3, no. 1, pp. 1-26, Mar. 1992.
[32] H. Fowler and W. Leland, Local Area Network Traffic
Characteristics, with Implications for Broadband Network Con-
gestion Management, IEEE J. Selected Areas in Comm., vol. 9, no. 7,
pp. 1139-1149, Sept. 1991.
[33] W. Leland, M. Taqqu, W. Willinger, and D. Wilson, On the Self-
Similar Nature of Ethernet Traffic, IEEE/ACM Trans. Networking,
vol. 2, no. 1, pp. 1-15, Feb. 1994.
[34] V. Paxson and S. Floyd, Wide Area Traffic: The Failure of Poisson
Modeling, IEEE/ACM Trans. Networking, vol. 3, no. 3, pp. 226-244,
July 1995.
[35] W.W.H. Chan and V. Leung, Performance Evaluations of Path
Optimization Schemes for Inter-Switch Handoffs in Wireless ATM
Networks, Wireless Networks, vol. 6, no. 4, pp. 251-262, July 2000.
NadjiaKarareceivedthePhDdegreein electrical
engineering from Ecole Polytechnique of Mon-
treal. She worked as a researcher and system
architect in the industry for more than 10 years.
She is an associate professor at the Institut
National de la Recherche Scientific-Energy,
Materials and Telecommunications (INRS-
EMT). Her research interests are in mobile
computing and networking, and are mainly fo-
cused on network architectures, services, and
applications suchas context-awareservices, multimediacommunications
in wireless networks, and Quality of Service architecture and manage-
ment. She is a member of the IEEE.
> For more information on this or any other computing topic,
please visit our Digital Library at www.computer.org/publications/dlib.
KARA: MOBILITY MANAGEMENT APPROACHES FOR MOBILE IP NETWORKS: PERFORMANCE COMPARISON AND USE... 1325