A Contribution to Fast Detection Movement
Javier Carmona-Murillo, José-Luis González-Sánchez
Department of Computing and Telematics System Engineering, University of Extremadura, Cáceres, Spain
Isaac Guerrero-Robledo
Department of Telecom Advances Services, Next Generation Intelligent Networks Group, Atos Origin, Madrid, Spain
Keywords: Mobile communications, movement detection, FDML3, handover, Mobile IPv6, OMNeT++.
Abstract: Nowadays, mobile communications face new challenges in its evolution: The convergence of wireless
cellular networking and TCP/IP architecture. In addition, Internet protocols do not support mobility, so
different mechanisms to offer seamless mobility have been proposed. The fourth generation (4G) IP-based
wireless networks have lead IP level the ideal candidate where mobility should be implemented. Mobile IP
is the protocol proposed for mobility management at the IP layer. Handover management is one of the most
critical phases of this protocol. The high delay of this phase is a limitation to seamless mobility. In this work
a detailed analysis about handover process is presented. Moreover, movement detection is a very costly
stage in handover mechanism so a new fast movement detection algorithm to improve this detection has
been developed. It is called FDML3 (Fast Detection Movement Layer 3). As the handover analysis as the
algorithm proposed has been carried out thanks to OMNeT++ simulator.
In recent years mobile communications have
changed the traditional way of Internet access.
Convergence between TCP/IP and wireless networks
is a challenge to achieve seamless mobility in
heterogeneous networks (Makaya, 2007). In this
sense, mobility management must be implemented
in a common level. IP is the best candidate to offer
this capacity (Abduljalil, 2007). However, IP
operation does not allow a node to move between
different networks without a connection disruption.
To solve this situation, IETF (Internet Engineering
Task Force) has designed Mobile IPv6 (hereafter
MIPv6) (Johnson, 2004). One of the most critical
phases in MIPv6 is the handover or handoff,
produced when a node moves to a new IPv6 subnet
while connection is still alive (Koodly, 2007).
In this work, we analyze the handover process
and a new L3 detection movement algorithm has
been developed (FDML3) to decrease the handover
delay. This work belongs to a research project called
Campus Ubicuo (Carmona-Murillo, 2007).
This article is organized as follows: Section 2
presents mobility problem in IP networks; section 3
is focused on handover analysis; the proposed
FMDL3 algorithm is explained in section 4; next,
simulation results are shown in section 5; finally,
conclusions are presented in section 6.
In general, a host in the Internet changes data with
other nodes thanks to TCP/IP architecture. These
protocols were designed for fixed hosts, which are
identified by an IP address.
Convergence towards “All-IP” architectures next
generation wireless networks has made Mobile IP
(Figure 1) the main solution to offer seamless
mobility in the Internet (Le, 2006). Some approaches
to reduce movement detection latency are based on
layer 2 information. These solutions are faster than
L3 ones but have an important drawback because
they restrict the movement among heterogeneous
networks due to L2 access technology dependence.
Carmona-Murillo J., González-Sánchez J. and Guerrero-Robledo I. (2008).
HANDOVER PERFORMANCE ANALYSIS IN MOBILE IPv6 - A Contribution to Fast Detection Movement.
In Proceedings of the International Conference on Wireless Information Networks and Systems, pages 78-81
DOI: 10.5220/0002023100780081
Figure 1: Mobile IPv6 entities.
Mobility protocols are designed to solve overhead,
packet loss and path recovery during handover.
Handover latency is defined as the interval starting
when the mobile node (MN) leaves the old access
medium until the communication is resumed
(Figure2). In this work a handover analysis has been
carried out, contrasting it with similar research in
this area (Cabellos-Aparicio, 2005).
Figure 2 shows handover latency components.
T1 is the L2 handover and represents 12% of the
total handover latency. T2 is the time spent by IPv6
to realize that it is attached to a new subnet and to
obtain a new IP (87%). Finally, MIPv6 operation is
carried out in T3 and is composed of the time that
the MN needs to announce its new location (1%).
Accordingly, the handover delay is given by (1):
= T
+ T
In this work, we focus our attention in L3
handover delay (2). In detail, this time results:
= T
+ T
where T
(3) and T
(4) are:
= T
+ T
= T
+ T
As Table 1 shows, T2 is the main responsible of
the high latency in handover process, so most of the
time (87%) is devoted to IPv6 tasks.
Table 1: Phases in MIPv6 handover.
Handover phases Time
T1 = T
T2 = T
T3 = T
Figure 2: Handover latency components.
Movement detection (MD) is a crucial task in
MIPv6 handover and is based on IPv6 Neighbour
Discovery (ND) (Narten, 1998). MD in MIPv6 is too
costly in spite of modifications performed by MIPv6
in ND process. The most important modification is
to allow sending Router Advertisements (RA) more
frequently than the 3 second established in standard.
MinRtrAdvInterval and MaxRtrAdvInterval can be
set up till 0,03 and 0,07 seconds. Some studies have
provided a mathematical analysis of MD in MIPv6
(Young-Hee, 2006), (Lee, 2004). In this work, we
propose a L3 fast detection movement mechanism
called FDML3. This algorithm starts from the
research developed in (Blefari-Melazzi, 2005). The
algorithm flowchart (Figure 3) is explained next.
1. A MN detects that an unsolicited RA has been
lost. This situation is known because of the
absence of a new unsolicited RA in an interval
equal to the interval option configured.
2. MN sends a Router Solicitation to the access
router to check the bidirectional reachability.
3. If a RS is not received in an interval between 0
second), is possible to suppose that the lost has
been caused due to a mobile node movement.
4. The MN tries to connect to a new access router
to complete the handover, listening RA
messages sent by routers periodically. The
HANDOVER PERFORMANCE ANALYSIS IN MOBILE IPv6 - A Contribution to Fast Detection Movement
network prefix obtained in a RA is used to
configure the new CoA. This new IP address is
registered in the Home Agent (HA) and CN.
OMNET++ simulation scene (Figure 4) is composed
of nine routers, one of them is the Home Agent; nine
wireless access points; a MN (client1); and a CN.
The node moves in a circular path across the nine
access points (generating eight L3 handovers).
The three tests that appear in this section are:
1. L2 vs. L3 handovers in MIPv6
2. Influence of unsolicited RA interval, given by
MaxRtrAdvInterval and MinRtrAdvInterval.
3. Comparison of MIPv6 handover depending on
MD algorithm used (FMDL3 evaluation).
For each test, the following information is presented:
Overall MIPv6 handover time (sec.).
Data lost percentage in transmission (%).
Figure 3: FDML3 flow chart.
5.1 Link-layer Triggers
Although our proposal consider network-layer the
place where implement mobility, is important to
compare the behaviour when link-layer information
is used. As we can see in Figure 5 and in Table 2,
the difference between both times is very large.
Nowadays, handovers use L2 information to achieve
a high performance. In this test, L2 handover time is
75% less than the L3 one. This test proves the
necessity of L3 handover delay improvement to
achieve mobility in heterogeneous networks.
Figure 4: Simulation scene.
5.2 Interval between Unsolicited RAs
Routers send unsolicited RAs to advertise its
presence to other nodes in an interval defined by
MaxRtrAdvInterval and MinRtrAdvInterval. MIPv6
modify the default values of these parameters to
allow fast movement detection in network layer.
Figure 6 shows L3 handover delay in four
configurations, where these two parameters change
its values. Obtained data are also shown in Table 3.
If routers send unsolicited RAs fast, the time needed
to detect the movement is shorter. However, a low
configuration of these parameters causes an extra
overload in the network. As in test before, packet
loss is lower according to the time of the process.
Table 2: Simulation data. L2 triggers.
L2 handover L3 handover
Hand. Time 0,58 2,3
Packet loss 1,71 2,77
Table 3: Simulation data. Unsolicited RA interval.
0,5–1,5 0,3-0,1 0,25-0,75 0,1-0,3 0,03-0,07
Hand. time 2,30 1,32 0,96 0,18 0,15
Packet loss 2,76 2,20 1,87 1,47 1,21
Figure 5: Handover delay with and without L2 triggers.
WINSYS 2008 - International Conference on Wireless Information Networks and Systems
Figure 6: Handover delay. Unsolicited RA interval.
5.3 FMDL3 Evaluation
This proof checks the behaviour of the proposed
FMDL3 algorithm comparing it versus the defined
in (Johnson, 2004). Figure 7 and Table 4 shows the
simulation results. When FDML3 is used, handover
delay is reduced in an average of 25,6 %.
With this new mechanism, if unsolicited RA
interval established is low, the handover delay
improvement is not so high; however a configuration
like this provokes a high amount of signalling traffic
in the network, so this configuration will not be
chosen usually. This means that FDML3 algorithm
will improve the delay of the overall handover
process in MIPv6 protocol, minimizing connection
disruption while the mobile node moves among
heterogeneous networks.
Table 4: Simulation data. FDML3 algorithm.
With FDML3 Without FDML3
Hand. time 1,63 2,19
Packet loss 3,62 4,04
Figure 7: Handover delay improvement with FDML3.
In this work a MIPv6 handover evaluation is
presented, checking each phase of the process. This
analysis has been carried out using OMNeT++
simulator. Obtained data shows that there is a phase
very costly in time terms (87% of the process). In
this stage movement detection is performed, so it is
a critical part of the process. Due to this limitation, a
new fast movement detection algorithm has been
developed: FDML3. With this algorithm, the overall
delay is improved up to 25%.
Although this research work reduces the
handover delay, there are other important sources of
delay in MIPv6 handover: Router advertisement,
duplicated address detection (DAD) and Binding
Update RTT. The study and improvement of these
topics is presented as future work.
Masaya, C., Pierre, S., 2007. An Architecture for Seamless
Mobility Support in IP-based Next-Generation
Wireless Networks. In IEEE Transactions on
vehicular technology.
Abduljalil, F. M., Bodhe, S. K., 2007. A Survey of
Integrating IP Mobility Protocols and Mobile Ad Hoc
Networks. In IEEE Communications Surveys &
Tutorials, vol. 9, no. 1, pp. 14-30.
Johnson, D., Perkins, C., Arkko, J., 2004. Mobility
Support in IPv6. IETF RFC 3775.
Koodli, R. S., Perkins, C., 2007. Mobile Internetworking
with IPv6. Wiley-Interscience.
Carmona-Murillo, J., González-Sánchez, J. L., Castro-
Ruiz, M., 2007. Technological Innovation in mobile
communications, developed with Free Software:
Campus Ubicuo. In UPGRADE. Dec. 2007.
Le, D., Fu, X., Hogrefe, D., 2006. A review of mobility
support paradigms for the Internet. In IEEE
Communications Surveys, vol. 8, no. 1, pp. 38-51.
Cabellos-Aparicio, A., Serral-Gracià, R., Jakab, L.,
Domingo-Pascual, J., 2005. Measurement based
analysis of the handover in a WLAN scenario. In
PAMW2005. ISBN 3-540-25520-6, Boston (USA).
Narten, T., Nordmark, E., Simpson, W., 1998. Neighbor
Discovery for IP version 6 (IPv6). IETF RFC 2461.
Young-Hee Han, Seung-Hee Hwang, 2006. Movement
Detection Analysis in Mobile IPv6. In IEEE
Communications Letters, vol. 10, no. 1, pp. 59-61.
Lee, J. S., Koh, S. J., Kim, S. H., 2004. Analysis of
Handoff Delay for Mobile IPv6. In VTC 2004-Fall,
vol. 4. ISBN 0-7803-8521-7, Los Angeles (USA).
Blefari-Melazzi, N., Femminella, M., Pugini, F., 2005. A
Layer 3 Movement Detection Algorithm Driving
Handovers in Mobile IPv6. In Wireless Netwoks. Vol.
11 (Springer), pp 223-233.
HANDOVER PERFORMANCE ANALYSIS IN MOBILE IPv6 - A Contribution to Fast Detection Movement