AN ANALYSIS OF THE FLOW-BASED FAST HANDOVER METHOD

FOR MOBILE IPV6 NETWORK

Jani Puttonen, Ari Viinikainen, Miska Sulander and Timo H

ainen

Department of Mathematical Information Technology

University of Jyv

askyl

40014 Jyv

askyl

a, FINLAND

Keywords:

Mobile IPv6, FFHMIPv6, handover, wireless.

Abstract:

Mobile IPv6 has been proposed by the IETF (Internet Engineering Task Force) to be the solution to mobility

management in IPv6 network. The work is now culminating to a standard status. But, one problem still

remaining is the length of the handover time, which might cause packet loss. Thus the handover time should

be as short as possible. Especially the real-time trafﬁc suffers from packet loss. Earlier we have introduced

a new method for faster handover process in Mobile IPv6 network called the Flow-based Fast Handover

Method for Mobile IPv6 (FFHMIPv6). FFHMIPv6 uses the ﬂow state information stored in the routers for

the fast redirection of the ﬂow. In this paper we compare the proposed FFHMIPv6 protocol to other methods

using both theoretical analysis and Network Simulator 2 (ns-2) simulations.

1 INTRODUCTION

The evolution of mobility in IPv6 networks (Johnson

et al., 2004) has been signiﬁcant when compared to

the Mobile IPv4 protocol (Perkins, 2002). Mobile

IPv6 (MIPv6) provides transparent routing of IPv6

packets to Mobile Nodes (MNs) from Correspon-

dent Nodes (CNs). The mobility is achieved by us-

ing a Home Agent (HA) and a local Care-of-Address

(CoA). Unfortunately, the minimization of the han-

dover time between two logical subnets is still un-

solved. Real-time applications, like VoIP, are intol-

erant to delay and jitter. On the other hand the han-

dover time causes packet loss, which might affect the

application in use (e.g. real-time multimedia).

During the past few years, several different propos-

als have been presented to decrease the handover de-

lay in Mobile IPv6 networks. But, for some applica-

tions, the handover delays still remain unacceptable.

Also, many of the proposals require substantial and

undesirable modiﬁcations to the access routers (ARs)

or the MN.

The use of a “virtual” HA, located closer to the

MN than the actual HA is presented in (Castel-

luccia, 2000) and it is developed further in (Soli-

man et al., 2004). The HMIPv6 method has been

used as a ground for other proposals (Ramjee et al.,

2002)(Thing et al., 2003). Another approach to de-

crease the handover delays is based on multicast

routing (Ernst et al., 2000). Some proposals con-

centrate on the modiﬁcation of the MN, instead of

the AR(s) (Omae et al., 2002)(Patanapongpibul and

Mapp, 2003). Recently, optimization of routing pro-

tocols have also been proposed to decrease the de-

lay in handovers, such as improvements to HMIPv6

(Hwang et al., 2003)(Vivaldi et al., 2003), requiring

additional resources from the network elements. In

(Daley et al., 2003) a method is presented, where a

small part of the overall delay in MIPv6 handover is

addressed. This method can be used in conjunction

with most of the methods.

One way to reduce the packet loss during the han-

dover is to use tunneling and redirection of the ﬂow

heading for the old CoA. In (Koodli, 2004) is pre-

sented the Fast Handovers in Mobile IPv6 (FHO)

method, in which the ﬂow is tunneled from the pre-

vious AR to the new AR during the handover. FHO

method also employs L2 triggers to start the handover

procedure earlier. In (Sulander et al., 2004) we in-

troduced a new method for faster handover in Mobile

IPv6 network called the Flow based Fast Handover

for Mobile IPv6 (FFHMIPv6). By using the trafﬁc

ﬂow information each trafﬁc ﬂow can be identiﬁed

and redirected to a new location. The redirection takes

place in the router where the old and the new trafﬁc

ﬂow crosses. The method makes possible the recep-

187

Puttonen J., Viinikainen A., Sulander M. and Hämäläinen T. (2004).

AN ANALYSIS OF THE FLOW-BASED FAST HANDOVER METHOD FOR MOBILE IPV6 NETWORK.

In Proceedings of the First International Conference on E-Business and Telecommunication Networks, pages 187-190

DOI: 10.5220/0001390401870190

 SciTePress

tion of packets simultaneously with the BU registra-

tion process, thus minimizing the delay and packet

loss experienced in the handover.

In addition to faster handover proposals several per-

formance studies have been published. In (Montavont

and Noel, 2002) the performance of the basic Mo-

bile IPv6 is analyzed. Under evaluation are the L2

and L3 handover latencies up to 4 mobile nodes. The

performance analysis of FHO have been performed in

(Torrent-Moreno et al., 2003). The number of MNs,

the handoff rate, the distance of the CNs and HA, ef-

fect on different applications etc. were analyzed using

ns-2 simulations. FHO was found to be more effec-

tive than MIPv6 to the point where the radio chan-

nel is congested because of the number of MNs and

the amount of trafﬁc. This analysis was expanded

in (Perez-Costa et al., 2003) to include Hierarchical

MIPv6 and the combination of the HMIPv6 and FHO.

The combination was found to have the shortest han-

dover delay and lowest packet loss.

The remainder of the paper is organized as fol-

lows. Section 2 presents the idea of the proposed

FFHMIPv6 method. The method is presented more

thoroughly in (Sulander et al., 2004). Analysis meth-

ods and the achieved results are presented in section

3. Finally, in section 4 we discuss conclusions and

future work.

2 FFHMIPv6 METHOD

In (Sulander et al., 2004) a Flow-based Fast Han-

dover for Mobile IPv6 (FFHMIPv6) method is pro-

posed for MIPv6 networks. It uses the ﬂow state in-

formation of the routers to locate the ﬂows heading

for the old CoA. These ﬂows are then encapsulated

and redirected to the new CoA during the BU process.

When the MN moves to a new logical subnet, it

receives a new CoA and registers it to the HA and

possibly to the CN(s) via BU process. In FFHMIPv6

method the Hop-by-Hop frame, including the the old

CoA, is added to the BU register message heading for

the HA.

Every router maintain the state information of the

ﬂows it receives, sends or routes. A Flow is deﬁned

by the source and destination addresses and the Flow

Label. In every router between the MN and the HA

the ﬂow state information and the old ﬂow informa-

tion from the Hop-by-Hop header is compared. If the

trafﬁc ﬂow is found, an IPv6 tunnel is established be-

tween the crossover router (CR) and the nCoA of the

MN and the trafﬁc ﬂow is redirected to the established

tunnel. Next, the Flow Path bit in the Hop-by-Hop

frame is set to one, so that the FFHMIPv6 process is

not performed again in another router. Finally the BU

message is forwarded towards the HA and in the next

hop the same procedure is repeated.

During the tunneling MN has had the time to reg-

ister the new CoA to the HA and CN(s). The FFH-

MIPv6 enables the receiving of the trafﬁc ﬂow si-

multaneously with the BU process; thus minimiz-

ing downstream packet loss. The FFHMIPv6 method

functions best as a micro mobility solution. The net-

work topologies are often built hierarchically so that

all of the domains ingress and egress trafﬁc pass a

same router (border router). Given this assumption

the crossover router would probably be found. If the

ﬂows are not found from the routers ﬂow state infor-

mation or the routers do not support FFHMIPv6, the

FFHMIPv6 functions just like MIPv6 and its BU pro-

cess.

3 ANALYSIS METHODS AND

RESULTS

We analyzed the FFHMIPv6 method theoretically and

with ns-2 (ns-2, 2004) simulations. The analysis with

both analysis methods is meant to be performed iden-

tically so that the results could be compared. Theoret-

ically we compare the basic MIPv6, HMIPv6, FHO

and FFHMIPv6 methods. The ns-2 simulations are

performed with MIPv6 and FFHMIPv6 methods. We

are working on the HMIPv6 and FHO implementa-

tions to ns-2.

The scenarios shown in Figure 1 are the same in

both analysis methods. The route optimization is not

used, so all the ﬂows from CNs are routed via HA.

The FFHMIPv6 method is therefore used to redirect

the ﬂow from the HA to the new CoA. The handover

delay is deﬁned to be the time from the ﬁrst BU regis-

ter message to the time when the MN is able to receive

the ﬂow. Packet loss is deﬁned to be the loss due one

speciﬁc handover. While MN is in the overlapping

area of the BSs, according to IEEE 802.11 WLAN

standard (IEEE 802.11, 1999), MN can receive and

send data only from one BS at a time.

3.1 Theoretical Analysis

In (Sulander et al., 2004) we compared theoretically

the handover delays of the MIPv6, HMIPv6 and FFH-

MIPv6. In this paper we extend this analysis to in-

clude the Fast Handovers for Mobile IPv6 (FHO)

(Koodli, 2004). We concentrate on the situation,

where the MN can not anticipate the shortly occurring

handover. So it can’t receive L2 triggers or predict the

next access router. This assumption has an effect on

the performance of the FHO method.

We use two sample scenarios which represent the

handover situation in the best (scenario 1) and the

worst case (scenario 2) for the FFHMIPv6 method

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

188

PAR

NAR

Movement

d20

Domain A

Scenario 1

Scenario 2

MAP

PAR

Domain A

Domain B

Movement

MAP

NAR

MAP

d20

Figure 1: The analysis scenarios

(Figure 1). In scenario 1 the MN changes access

router within the same MAP domain (MAP exists

only in HMIPv6 network) and the access routers

(PAR and NAR) are connected to the same router (the

cross-over router in FFHMIPv6). In scenario 2, the

MN changes it’s MAP domain and there does not ex-

ist a cross-over router for FFHMIPv6.

There exists mainly two factors which affect the

MN’s MIPv6 handover delay – the number of regis-

trations and the time to accomplish one registration.

In the case, where the route optimization is not in use,

the handover delay consists of the BU process to the

HA. There are also other factors like movement de-

tection, acquiring new CoA, the processing delay in

the routers inﬂicted by the Hop-by-Hop header and

the ﬂow state information procedures. The processing

delays are to be analyzed in future work. Acquiring

the new CoA with stateless address autoconﬁguration

(including duplicate address detection (DAD)) causes

a lot of delay, but it has similar effect on every han-

dover method. Because it does not affect the ratios of

the handover delays, they were not taken into consid-

eration.

We assume that the links in Figure 1 have delays

d1 = 1ms, d2 = 2ms and d20 = 20ms. The delays

have been selected to represent network where access

routers are near, but HA and CNs can be located quite

far. The values are not selected to improve the perfor-

mance of the proposed method.

Figure 2: Theoretical analysis results in the good and bad

scenarios

MIPv6 requires the handover time of the BU pro-

cess to the HA. The BU process is similar in both sce-

Table 1: The simulation results

Scenario 1 Scenario 2

MIPv6 FFHMIPv6 MIPv6 FFHMIPv6

Delay

(ms) 58.8 13.8 65.2 65.2

Loss

(pkts) 5 1 6 6

narios. In scenario 1 HMIPv6 needs only the BU reg-

istration to the MAP because the MN moves within

its MAP domain. In scenario 2 the MN must per-

form the BU process to the new MAP and HA. Be-

cause the occurring handover can not be predicted,

FHO requires the signaling to the previous AR to es-

tablish the tunnel to the new AR. This assumption de-

grades the performance of the FHO method signiﬁ-

cantly. FFHMIPv6 requires only the BU message to

the crossover router in the scenario 1. In scenario 2

the crossover router is not found, so the FFHMIPv6

method is in practice functioning as MIPv6. Figure 2

presents the results of the theoretical analysis.

3.2 Simulation studies

Mobile IPv6 extensions for Network Simulator 2 (ns-

2, 2004) have been developed by Motorola Labs

Paris in collaboration with Inria Planete team at Inria

one-Alpes. The extension called Mobiwan (Ernst,

2002) works in ns version 2.1b6. We implemented the

FFHMIPv6 protocol to this environment.

The link bandwidths were chosen so that they do

not affect the simulation results. CN sends constant

bit rate (CBR) trafﬁc to the MN during the han-

dover. Because route optimization is not used the

ﬂow is routed via HA. CBR trafﬁc’s packet size is

500 bytes and packet sending interval 0.05 seconds

(10kbps). MN originated in the PAR area moves at

constant speed of 15 m/s towards the NAR and does

the L3 handover with MIPv6 and FFHMIPv6 han-

dover methods. The handover delay and resulting

packet loss is calculated in both situations.

Results are presented in Table 1. In scenario 1

the FFHMIPv6 handover time is much shorter than

in MIPv6, because the BU register message to the

crossover router R1 is enough. R1 tunnels the CBR

trafﬁc to the new location of the MN. The packet

loss is also very self-explanatory, because the trafﬁc

MN receives is constant bit rate sent at the interval of

10 ms. In scenario 2 the FFHMIPv6 is functioning

as MIPv6, because the crossover router is not found.

The results achieved by the ns-2 simulations are much

similar as expected due to theoretical analysis, thus

simular results are expected also to the HMIPv6 and

FHO handover methods.

AN ANALYSIS OF THE FLOW-BASED FAST HANDOVER METHOD FOR MOBILE IPV6 NETWORK

189

4 CONCLUSIONS

In (Sulander et al., 2004) we presented a new Flow-

based Fast Handover method for Mobile IPv6 net-

work. In this paper we expanded the performance

analysis of the FFHMIPv6 method. The proposal was

compared theoretically and with ns-2 simulations to

other methods.

Theoretical analysis were made with statical link

delays and calculating the required signaling. FFH-

MIPv6 was implemented to ns-2 and it was com-

pared with simulations to the basic Mobile IPv6. Both

the theoretical analysis and simulations were done in

the so-called good and bad case handover scenarios

and we found the handover delay to be signiﬁcantly

shorter than other handover methods under the given

assumptions.

The future work includes the expansion of the FFH-

MIPv6 performance analysis. We are implementing

the HMIPv6 and FHO handover methods to ns-2. In

addition the simulations are going to be performed to

include the variables of number of MNs and the han-

dover rate. MNs sporadic movement and different ap-

plications are also to be considered. These bring us

closer to more realistic situation. Also the required

processing time or the processing load to the core

routers is brought as a new parameter besides the han-

dover delay and the packet loss.

REFERENCES

Castelluccia, C. (2000). HMIPv6: A hierarchical mobile

IPv6 proposal. ACM Mobile Computing and Commu-

nications Review, 4(1):48–59.

Daley, G., Pentland, B., and Nelson, R. (2003). Effects

of fast router advertisement on mobile IPv6 han-

dovers. In Proceedings of the Eighth IEEE Interna-

tional Symposium on Computers and Communication

2003 (ISCC 2003), pages 557–562.

Ernst, T. (2002). Mobiwan: NS-2 extensions to

study mobility in wide-area IPv6 networks.

http://www.inrialpes.fr/planete/mobiwan/.

Ernst, T., Castelluccia, C., and Lach, H.-Y. (2000). Ex-

tending mobile-IPv6 with multicast to support mobile

networks in IPv6. In Proceedings of the 1st European

Conference on Universal Multiservice Networks 2000

(ECUMN 2000), pages 114–121.

Hwang, S.-H., Lee, B.-K., Han, Y.-H., and Hwang, C.-S.

(2003). An adaptive hierarchical mobile IPv6 with

route optimization. In Proceedings of the 57th IEEE

Semiannual Vehicular Technology Conference 2003

(VTC’S03), volume 3, pages 1502–1506.

IEEE 802.11 (1999). Wireless LAN medium access con-

trol (MAC) and physical layer (PHY) speciﬁcations.

IEEE Standard 802.11.

Johnson, D., Perkins, C., and Arkko, J. (2004). Mobility

support in IPv6. IETF Draft. work in progress.

Koodli, R. (2004). Fast handovers for mobile IPv6. IETF

Draft. work in progress.

Montavont, N. and Noel, T. (2002). Handover management

for mobile nodes in IPv6 networks. IEEE Communi-

cations Magazine, 40:38–43.

ns-2 (2004). The UCB/LBNL/VINT network simulator –

ns (version 2). http://www.isi.edu/nsnam/ns/.

Omae, K., Ikeda, T., Inoue, M., Okajima, I., and Umeda,

N. (2002). Mobile node extension employing buffer-

ing function to improve handoff performance. In Pro-

ceedings of the The 5th International Symposium on

Wireless Personal Multimedia Communications 2002,

volume 1, pages 62–66.

Patanapongpibul, L. and Mapp, G. (2003). A client-based

handoff mechanism for mobile IPv6 wireless net-

works. In Proceedings of the Eighth IEEE Interna-

tional Symposium on Computers and Communication

2003 (ISCC 2003), pages 563–568.

Perez-Costa, X., Torrent-Moreno, M., and Hartenstein, H.

(2003). A performance comparison of mobile IPv6,

hierarchical mobile IPv6, fast handovers for mobile

IPv6 and their combination. SIGMOBILE Mob. Com-

put. Commun. Rev., 7(4):5–19.

Perkins, C. (2002). IP mobility support for IPv4. IETF

RFC 3344.

Ramjee, R., Varadhan, K., Salgarelli, L., Thuel, S., Wang,

S.-Y., and Porta, T. L. (2002). HAWAII: a domain-

based approach for supporting mobility in wide-area

wireless networks. IEEE/ACM Transactions on Net-

working, 10:396–410.

Soliman, H., Castelluccia, C., El-Malki, K., and Bellier, L.

(2004). Hierarchial mobile IPv6 mobility manage-

ment (HMIPv6). IETF Draft. work in progress.

Sulander, M., Hamalainen, T., Viinikainen, A., and Put-

tonen, J. (2004). Flow-based fast handover method

for mobile IPv6 network. In Proceedings of the 59th

Semiannual IEEE Vehicular Technology Conference

(VTC’S04).

Thing, V., Lee, H., and Xu, Y. (2003). Performance evalu-

ation of hop-by-hop local mobility agents probing for

mobile IPv6. In Proceedings of the Eighth IEEE In-

ternational Symposium on Computers and Communi-

cation 2003 (ISCC 2003), pages 576–581.

Torrent-Moreno, M., Perez-Costa, X., and Sallent-Ribes,

S. (2003). A performance study of fast handovers

for mobile IPv6. In Proceedings of the 28th Annual

IEEE International Conference on Local Computer

Networks, pages 89–98.

Vivaldi, I., Ali, B., Habaebi, H., Prakash, V., and Sali, A.

(2003). Routing scheme for macro mobility handover

in hierarchical mobile IPv6 network. In Proceedings

of the 4th National Conference on Telecommunication

Technology 2003 (NCTT 2003), pages 88–92.

ICETE 2004 - WIRELESS COMMUNICATION SYSTEMS AND NETWORKS

190