A Novel Approach For Efficient Resource Consumption In GPS-Based Mobile IPv6

Wireless LAN

Amirhosein Moravejosharieh

Department of Computer System

and Technology

University of Malaya

Kuala Lumpur, Malaysia

Amirhosein.moravej@gmail.com

Rosli Salleh

Department of Computer System and Technology

University of Malaya

Kuala Lumpur, Malaysia

rosli_salleh@um.edu.my

Hero Modares

Department of Computer System

and Technology

University of Malaya

Kuala Lumpur, Malaysia

hero.modares@gmail.com

Abstract—Seamless handover has become an important

challenging issue among research groups over the past 5 years.

Researchers and scientists have been making efforts to provide

robust and integrated connectivity while the mobile node (MN)

moves between networks. Running certain procedures such as

movement detection (MD), neighbor discovery (ND), duplicate

address detection (DAD) and sending binding updates (BUs)

are needed in order to achieve complete handover procedure

which results in handover latency and eventually packet loss.

Recently, scientists have found that by utilizing information

provided by the Global Positioning System (GPS) device, it is

possible to reduce the amount of handover latency by

decreasing the amount of time consumed by MD, ND and DAD

procedures. In this paper we are going to investigate the

proposed GPS models and their effects on the handover procedure in the MIPv6 protocol.

Keywords: GPS; mobile IPv6; fast handover; handover

latency; seamless handover

I. INTRODUCTION

Mobile IPv6 provides the ability for hosts to benefit from travelling between points of attachment while still connected to the Internet. Changing points of attachment, which is known as handover procedure, requires running some sub-procedures like movement detection (MD), neighbor discovery (ND), duplicate address detection (DAD) and sending binding updates to the home agent and correspondent node. The mentioned sub-procedures consume time within which packets destined to the mobile node (MN) may not be delivered completely resulting in packet loss which is an important challenging issue when real-time applications such as VoIP or video conference are being used. Thus, in order to overcome this issue researchers have been trying to improve the handover latency by proposing various schemes and models. In this paper, different mobile IPv6 protocols are discussed along with some specific GPS techniques, in order to improve the performance of handover latency and the effects on other network parameters such as bandwidth consumption and resource usage. Recently researchers have figured out that coordinate information can help MN detect its movement in addition to determining the next neighbor network that MN is going to travel to. This

paper is divided into two compartments. The first part investigates the Fast Handover MIPv6 protocol (FMIPv6) which is standardized by IETF research groups [1] and the latest related handover latency improvements. The second part is about the GPS model proposed in order to reduce the handover latency by deploying GPS information. Although the GPS-aided MIPv6 protocol has decreased the amount of latency, some other issues such as high bandwidth consumption and resource usage have arisen. In this paper we propose a new model which significantly reduces the amount of bandwidth consumption and resource usage.

II. RELATED WORKS

A. Fast Handover for mobile IPv6

As previously mentioned, there are a lot of implementations available for Mobile IPv6, some of which have been successful, while some concepts such as seamless handover, security and bandwidth consumption are still issues that researchers are working on and trying to improve. One of the areas that researchers have focused on recently is Fast Handover for Mobile IPv6 which focuses on decreasing packet loss during the handover process, in other words, it aims to decrease handover latency [2].

1) Predictive FMIPv6 In FMIPv6, when the MN detects its movement, it will

send the “Router Solicitation for Proxy Advertisement” (RtSolPr) message. Before sending this message it first gets some information about the neighboring APs and their subnet. The MN does this by listening to the link layer and running a procedure called link layer scanning. So, the default router replies to the message by sending the Proxy Router Advertisement (PrRtAdv) message which contains a set of ICMPv6 options that gives some information about the Access Router`s (AR) link layer address, AP`s link layer access and AR`s subnet prefix and prefix length. After receiving the PrRtAdv message, MN will send the Fast Binding Update (FBU) message to its current access router that is called Previous Access Router (PAR) and sends its CoA to the current network and the desired Neighbor Access Router (NAR). In FMIPv6, the handover latency involved in the BU`s process is decreased by establishing a bi-directional tunnel [3]. When the PAR receives this message from the

2011 Third International Conference on Computational Intelligence, Modelling & Simulation

978-0-7695-4562-2/11 $26.00 © 2011 IEEE

DOI 10.1109/CIMSim.2011.61

303

MN, it will start Handover Initiation (HI) with the mentioned NAR. The HI message contains some information about MN and link layer address. When NAR accepts this connection, a Handover Acknowledgement (HACK) message will be sent from NAR to PAR that contains some additional information needed to switch the links. When the PAR receives the HACK message it will send the needed information to MN by sending a message called Fast Binding Acknowledgment (FBACK). Now the MN is ready to go to another network and switch the links. The moment it reaches the new network, it will send a Fast Neighbor Advertisement (FNA) message to NAR so that NAR will update its cache entries and this is the point where handover is completed. Fig 1 indicates the basic functionality of predictive FMIPv6.

Figure 1. Predictive FMIPv6

2) Reactive FMIPv6 There is another scenario in which the FMIPv6 protocol

has defined another method for the handover, called reactive

handover. In this method the MN is not able to predict the

handover so it has to do the handover procedure while it is

in progress. In this case, after completing layer 2 handover, the MN will send a message to NAR which is the FBU

encapsulated in FNA message. By receiving this message

with NAR, the NAR will forward the FBU message to PAR,

then both PAR and NAR will start to exchange HI and

HACK messages, so that the MN can switch the links [4].

Fig 2 depicts the basic functionality of reactive FMIPv6.

In order to reduce packet loss, a new scheme is proposed

in literature [5], in which when MN is moving to the

neighbor network and still has its connection to the current

network, in addition to performing the FMIPv6 handover as mentioned above, PAR and NAR can establish a bi-

directional tunnel so that when the MN moves to the new

network the old AR can forward the buffered traffic to the

new one. By utilizing a combination of Proactive Binding

for FMIPv6 (PB-FMIPv6) [6] and Enhanced Route

Optimization (ERO) [7], the amount of handover latency

can be reduced significantly [8]. With PB-FMIPv6 there is

no DAD procedure taken by MN, the NAR would do the

DAD procedure by receiving the HI message from PAR that

contains the new Care of Address (CoA) of the MN, thus

before MN enters into the new network, the DAD can

already be checked by NAR on behalf of MN. By utilizing

the ERO technique, the MN is able to establish the

connection with CNs before leaving its home network and

by some special technique it can maintain the security of that connection.

Figure 2. Reactive FMIPv6

Some procedures have been proposed in order to reduce

the amount of time consumed by the DAD procedure and to

try to decrease the probability of being duplicated down to

near 0% [9]. Some others proposed another method in

which NAR manages and takes care of the creation of new addresses [10] by changing the network to cell clusters with

unique IDs.

Literature [11] states that by adding some extra

information into ICMPv6 packets and running certain

procedures, the amount of handover latency due to neighbor

discovery can be reduced. In this case, some extra

information necessary for seamless handover is added to

ICMPv6 packets in both predictive and reactive FMIPv6

[12] (e.g. FBU message includes Context Transfer Activate

Request (CTAR) message and the HI message exchanged

between ARs contains Context Transfer Data (CTD) message in predictive mode and another sequence of IPv6

messages for reactive).

In all the aforementioned schemes and models, the most

important issue that researchers are interested in is having

the handover latency reduced by establishing a bi-

directional tunnel for each MN that intends to move to a

neighbor network, buffering packets, expanding datagram to

carry out extra information, changing the network to cell

clusters that increases the cost of network structure and in

some cases all the burdens are located on Access Points and

Access Routers. All these proposed techniques and schemes

result in noticeable over usage of resources and significant bandwidth consumption. The next section discusses

utilizing new models in which the MN can benefit from

some location information provided by a GPS device, so

304

that the MN can perform the handover procedure with

minimum handover latency.

B. GPS models

As previously mentioned, movement detection and

neighbor discovery are related to MN location. Thus,

knowing the destined neighbor network can save huge

amounts of bandwidth consumption and resources used to

perform the handover procedure properly. In mobile IPv6,

most of the researchers have been trying to reduce the

handover latency without considering exactly which

neighbor network the MN intends to travel to.

The GPS model proposed in literature [13] reduced the unnecessary handover that results in increasing handover

procedure performance by combining the location

information and Signal to Noise Ratio (SNR). The proposed

model may not work properly all the time because one of

the factors involved in this combination is SNR value which

may not provide the right answer all the time.

In literature [14], an MN periodically sends its

coordinate information to the AP and AP replies appropriate

information based on MN location. Assuming there are

hundreds of MNs moving within the home network and just

a few intend to travel to a neighbor network, all are consuming bandwidth because of GPS information

exchange between MN and AP. So it seems this approach

does not focus on bandwidth consumption either. However,

a good point in this model is that handover burdens are

distributed to both MN and AP.

Other researchers have created a new GPS model in

which the MN has to calculate its speed and predict its

future positions, and by running a huge number of processes

it can predict the future intended neighbor network. It seems

this model not only does not consider the amount of

resource usage, but also the MN has to incur tremendous

burdens [15]. Literature [16] proposed a new GPS model in which two

techniques are deployed. The first is the GPS technique that

retrieves GPS information and calculates an angle in order

to find out the direction of MN movement, and the other is

utilizing software called “inSSIDer” in order to obtain the

Receive Signal Strength Indicator (RSSI) value. Both these

calculated values are inserted into fuzzy logic system and

the result is the priority cost for each AP in order to choose

the right AP and to avoid starting the handover procedure

with the wrong AP. This technique reduces bandwidth

consumption but the amount of resource usage is still significantly high due to processing huge amounts of data as

well as the possibility of providing the wrong answer

because of signal strength utilization as one of the effective

factors inserted as an input to that fuzzy logic system.

In order to provide a trade-off between seamless

handover and bandwidth/resource consumption we propose

a new GPS model that provides handover pre-configuration

information for MN and reduces the total bandwidth

consumption and resource usage due to calculating and

exchanging GPS information messages simultaneously.

III. PROPOSED GPS MODEL

Fig 3 indicates the simple structure of two neighbor

networks and an MN which is equipped with a GPS device, and also the calculated angle that shows the direction of

MN.

Figure 3. Neighbor networks structure

In our proposed model, an MN detects its movement by

calculating the distance between itself and AP. This

parameter is known as “d” and can be calculated by the

following mathematical formula (1) in order to determine if

it has either exceeded the preconfigured threshold or not:

d = (X1 – X2)2 + (Y1-Y2)

2 (1)

When an MN finds out that it has exceeded the

preconfigured threshold it proceeds to calculate the direction of its movement and sends out the current location and

calculated angle to the AP. Eventually, the AP will reply

with the appropriate commands and information needed for

starting the handover procedure.

IV. RESULTS AND DISCUSSION

By utilizing this model, there is no need for measuring

either RSSI value or the fuzzy logic system in order to

prevent starting the handover procedure with the wrong AP.

This means that the total resources consumed by MN are

reduced significantly in addition to decreasing the burden on

MN. In our proposed GPS model, effort is made to remove

the wrong effects of RSSI value, because sometimes an MN may be behind huge objects such as large buildings or face

some destructive noise that results in producing wrong

values. The number of processes compared to previous GPS

models not only is less, but also burdens due to calculating

the next neighbor are more evenly distributed between MNs

and APs, whereas the burden was only on MN side in most

of the previously proposed GPS models. This model also

reduces bandwidth consumption compared with the

proposed models in literature [14].

Fig 4 represents the proposed model process diagram for

movement detection and neighbor discovery processes that

305

MN performs while it moves within its current connected

network. As shown in Fig 4, there are no complicated

mathematical computations.

The proposed model can be considered as distributive

model. In this model, total amount of burden due to

providing those pre-configuration settings in order to have

seamless handover is distributed among network entities.

The signal strength information is not required in our

proposed model because the information needed to start

seamless handover will be sent by AP at the appropriate

time. As an example, following equations illustrate total

amount of resources consumed by network entities in our

proposed model and the model proposed in literature [16]:

RTotal (proposed model) = RD+RA+RKAP (2)

RTotal (literature [16]) = RD+RA+ RRSSI +RFLS+ RKMN (3)

where RD and RA are the amount of resources consumed by

MN in order to calculate the distance and angle respectively.

RKAP and RKMN are representatives of the amount of

resources consumed by AP and MN respectively in order to

maintain and process the information related to neighbor

networks. It can be assumed that RKAP and RKMN are equal

and the difference is negligible. RFLS stands for the amount of resources consumed by the fuzzy logic system in order to

provide a specific cost for each neighbor AP. By comparing

equation (2) and (3), our proposed model is clearly justified

in terms of resource consumption

The following table (Table 1) explains the strengths and

weaknesses of the proposed GPS models. Table 1

Investigated

issues FMIPv6

GPS models

Literature

[13]

Literature

[14] Literature

[15] Literature

[16]

Proposed

GPS

model

Packet loss

rate high low low Low low low

Burden on

MN normal high high High high low

Burden on

AP high normal low Low low normal

Bandwidth

consumption high high low Low low low

Resource

consumption high high high High high low

The above table describes major issues investigated in

this paper. As previously explained, the numerous burdens

caused by calculation and processes for finding the intended

neighbor network are reduced and evenly distributed

between MN and AP. The other point of strength is related

to resources consumed by MN in order to determine the

next neighbor networks.

V. CONCLUSION

Recently, researchers have been trying to reduce the

amount of handover latency by utilizing location

information provided by GPS devices in order to remove

time consuming processes related to the handover

procedure. In this paper, we described the main issues in

FMIPv6 handover and reasons that guide us to deploy new

GPS models. Although the handover latency is improved by

the new GPS model, attention is still needed to other issues

such as bandwidth consumption and resource usage which

are noticeably high in previous proposed GPS models. Here we have attempted to create a new GPS model in order to

provide acceptable balance and trade-off between resources

consumed by network entities and providing seamless

handover.

REFERENCES

[1] R. Koodli, "Fast {H} andovers for {M} obile {IP} v6," 2005.

[2] R. Li, et al., "An enhanced fast handover with low latency for

mobile IPv6," Wireless Communications, IEEE Transactions

on, vol. 7, pp. 334-342, 2008.

[3] A. Cabellos-Aparicio, et al., "Evaluation of the fast handover

implementation for mobile IPv6 in a real testbed," Operations

and Management in IP-Based Networks, pp. 181-190, 2005.

[4] E. Ivov and T. Noel, "An experimental performance evaluation

of the IETF FMIPv6 protocol over IEEE 802.11 WLANs,"

2006, pp. 568-574.

[5] M. Alnas, et al., "Performance evaluation of mobile IPv6 fast

handover," 2010, pp. 85-89.

[6] F. Z. Yousaf and C. Wietfeld, "Proactive Bindings for FMIPv6,"

draftyousaf-ietf-mipshop-pbfmipv6-00. txt, Internet Draft, IETF,

2007.

Start movement

Retrieve GPS information

Calculate d parameter

Interval time

Exceed

preconfigured

threshold

Calculate angle

Send both values to AP

YES

NO

Figure 4. Process diagram

306

[7] J. Arkko, et al., "Enhanced route optimization for mobile IPv6,"

draft-ietf-mipshop-cga-cba-03. txt, 2007.

[8] J. Espi, et al., "Proactive Route Optimization for Fast Mobile

IPv6," 2010, pp. 1-5.

[9] P. Pongpaibool, et al., "Fast duplicate address detection for

mobile IPv6," 2008, pp. 224-229.

[10] J. S. Lee, et al., "Fast handover scheme using temporary CoA in

Mobile WiMAX systems," 2009, pp. 1772-1776.

[11] R. L. Aguiar, "Some comments on hourglasses," ACM

SIGCOMM Computer Communication Review, vol. 38, pp. 69-

72, 2008.

[12] R. Farahbakhsh, "Smooth handover by synchronizing context

transfer protocol and fast mobile IPv6," 2010, pp. 1-5.

[13] A. Dutta, et al., "FRAMEWORK OF MEDIA-INDEPENDENT

PRE-AUTHENTICATION IMPROVEMENTS," ed: WO

Patent WO/2007/007,914, 2007.

[14] J. Montavont, et al., "Analysis of Mobile IPv6 Handover

Optimizations and Their Impact on Real-Time

Communication," 2007, pp. 3244-3249.

[15] H. Hamad and S. Elkourd, "New approach to eliminate the

latency in handoff schema for mobile node," Journal Media and

Communication Studies Vol, vol. 2, pp. 054-061, 2010.

[16] A. S. Sadiq, et al., "Mobility and Signal Strength-Aware

Handover Decision in Mobile IPv6 based Wireless LAN,"

Proceedings of the International MultiConference of Engineers

and Computer Scientists, vol. 1, 2011.

307