A Mobile IPv6 based Distributed Mobility Management Mechanism of Mobile Internet

Abstract

A flatter architecture is one of the trends of mobile Internet. Traditional centralized mobility management mechanism faces the challenges such as scalability and UE reachability. A MIPv6 based distributed mobility management mechanism is proposed in this paper. Some important network entities and signaling procedures are defined. UE reachability is also considered in this paper through extension to DNS servers. Simulation results show that the proposed approach can overcome the scalability problem of the centralized scheme.

Full text

Physics Procedia 25 ( 2012 ) 2249 – 2256
1875-3892 © 2012 Published by Elsevier B.V. Selection and/or peer-review under responsibility of Garry Lee
doi: 10.1016/j.phpro.2012.03.378
2012 International Conference on Solid State Devices and Materials Science
A Mobile IPv6 based Distributed Mobility Management
Mechanism of Mobile Internet*
Shi Yan1, Cheng Jiayin2, Chen Shanzhi3
1State Key Laboratory of Networking and Switching Technology
Beijing University of Posts and Telecommunications, Beijing, China
2International School, Beijing University of Posts and Telecommunications, Beijing, China
3State key Lab of Wireless Mobile Communications
China Academy of Telecommunication Technology, Beijing, China
Abstract
A flatter architecture is one of the trends of mobile Internet. Traditional centralized mobility management mechanism
faces the challenges such as scalability and UE reachability. A MIPv6 based distributed mobility management
mechanism is proposed in this paper. Some important network entities and signaling procedures are defined. UE
reachability is also considered in this paper through extension to DNS servers. Simulation results show that the
proposed approach can overcome the scalability problem of the centralized scheme.
© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [name organizer]
Keywords: mobility management, distributed, MIPv6, mobile Internet.
1. Introduction
With the dramatic growth of mobile terminals with rich capabilities, more and more users access
Internet through their mobile terminals. It is predicted that the mobile Internet user number will exceed
that of fixed Internet in four years [1]. The flat architecture is the necessary trend of mobile Internet for
reducing the cost and improving performance [2][3].
Mobility management is one of the key technologies responsible for maintaining communication
*This work is partially supported by Major National Science and Technology Special Project (No. 2009ZX003004-001,
2010ZX03005-002-02), the Fundamental Research Funds for the Central Universities (No. 2009RC0503).
Available online at www.sciencedirect.com
© 2012 Published by Elsevier B.V. Selection and/or peer-review under responsibility of Garry Lee
Open access under CC BY-NC-ND license.
Open access under CC BY-NC-ND license.

2250 Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256
continuity. Under the new trend of flat architecture, mobility management in mobile Internet faces some
important challenges. Firstly, it will incur more frequent handoff and will decrease the service experience
thereby. Secondly, it should support the heterogeneous wireless access environments. Lastly, UE global
reachability is necessary to support some new applications. On the other hand, Current IP mobility
management protocols are derived from Mobile IP[4] principles. They follow the centralized architecture
which means that there is always a centralized anchoring point in current mobility protocols, e.g., HA in
MIPv6/DSMIPv6, LMA in PMIPv6. The centralized approach fits well for the centralized network
architecture. However, the network is evolving towards to flatter architecture and the traditional mobility
management protocols can’t meet the new requirements and faces scalability problem. Therefore, the
distributed mechanism is the more reasonable way for the mobility management in the flatter network
architecture.
Some efforts have been done about the distributed mobility management mechanism. IETF discussed
the requirements and possible thinking of distributed mobility management in [5]. But this draft is still in
very early stage without any detailed technical solutions. [6][7] are distributed mobility management
schemes but are proposed for mobile Ad Hoc network and mobile sensor network. Detailed comparison
and simulation of centralized mobility management and distributed mobility management are given in [8],
which serves as a proof for distributed mobility management well adapted to flat network architecture. A
distributed approach for 3G SAE core network has been proposed in [9] distributing a MIP HA function
between several mobility agents. [10] proposed a distributed and dynamic mobility scheme based on the
principle that the mobility path should benefit from being managed and anchored as close as possible to
MN.
In this paper, a MIPv6 based distributed mobility management mechanism is proposed. The traditional
centralized HA are distributed across DMAs (Distributed Mobility Agent). The mobility management
related binding data is defined accordingly. Major signaling procedures such as attachment, location
update, paging and handoff are also introduced. Simulation results show that the proposed approach can
overcome the scalability problem of the centralized scheme. UE reachability is also considered in this
paper through extension to DNS servers. The rest of the paper is organized as follows. Section II
introduces the system architecture and network entities. Major signaling procedures such as attachment,
location update, paging and handoff are illustrated in Section III. Section IV describes the simulations and
results. And then the paper concludes in Section V.
2. System Architecture
A. Network Architecture
The network architecture of the distributed mobility management mechanism is shown in Fig. 1. It
follows the definition of Home Network/Foreign Network in MIPv6. In the different domains in Fig. 1,
one is home network and the others are foreign networks.

Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256 2251
Fig. 1 Network Architecture.
As shown in Fig. 1, MN is the mobile node moving from one network domain to another. CN
(Correspondent Node) is the peer with which MN is communicating and it may be either mobile or
stationary. Important network entities involved in this architecture include DMA (Distributed Mobility
Agent), AR (Access Router) and DNS server. DMA is responsible for storing mobility management
related information and routing data packet. AR handles MN’s attachment to the Internet. DNS server
acts as traditional DNS servers mapping between FQDN (Fully Qualified Domain Name) to IP address
and is extended to supporting UE reachability.
B. Network Entities
DMAs, ARs and DNS servers are the key network entities in the network architecture. Some
extensions are made to them to realize the distributed mobility management mechanism.
DMAs are distributed symmetrical entities in each domain. This paper uses H-DMA/F-DMA for the
DMAs located at home network/foreign network. DMAs act as the mobile agents maintaining mobility
management related binding information of registered mobile nodes. The storing and retrieving of the
information is based on DHT methods. Mobility management related information stored in DMA includes:
1) Intra-domain list: Intra-domain list maintains the information of registered mobile nodes which
move within the specific network domain where this DMA located. To be consistent with DHT storing
and accessing of data, the data contained in intra-domain list will be formatted as: <Hash (HoA), AR_IP,
MN_IP>.
2) Inter-domain list: Inter-domain list records the information of registered mobile node when it
moves to foreign network. To keep consistence with DHT methods of data storing and retrieving, data
stored in inter-domain list is in the form: <Hash (HoA), F_DMA_IP>, where HoA is the home address of
mobile node, and F-DMA_IP is the IP address of related foreign DMA.
AR provides network access, IP address allocation, packet forwarding functions to the mobile nodes.
Additionally, AR is the end of tunnel towards MN and therefore the last one hop to MN. AR maintains
the binding as of <HoA, IP>. HoA is the Home Address of MN, and IP is the current IP address assigned
by AR.
It is noted that DMA and AR are the functional entities in this system and some DMAs can also
implement the functions of AR. The distributed DMAs can be considered as an overlay structure over the
network.
DNS server is extended for maintaining the mapping between MN’s application layer identifier (e.g.
FQDN) and HoA. Such an extension is necessary for supporting the global reachability of MN. That is,

2252 Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256
given the FQDN of MN, CN queries DNS Server to get the home address of MN.
3. The Distributed Mobility Management Mechanism
Based on the above network architecture, the functional entities and mobility management related
binding information, the proposed distributed mobility management mechanism can be described by the
major signaling procedures including attachment, location update, paging and handoff.
A. Attachment Procedure
Attachment procedure is initiated when a MN accesses to the system for the first time. MN sends
Attach Request with its FQDN. The subsequent operations include IP address assignment by AR,
mobility management information update in DMA through DHT methods and MN’s DNS data
registration in DNS server. The signaling procedure for attachment can be described as in Fig. 2.
B. Location Update Procedure
Location update procedure is initiated when the attachment point of MN is changed. If the new
attachment point is in the same domain of the old one, it is intra-domain location update as shown in Fig.
3(a). Otherwise, it is inter-domain location update as shown in Fig. 3(b).
In the intra-domain location update, the mobility management information in intra-domain list
maintained by DMA should be updated. Additionally, the binding data in the previous AR should be
deleted.
In the inter-domain location update, besides the similar message exchanging in the intra-domain
location update, entry should be added to the intra-domain list in F-DMA. An Binding Update message
should be sent to update the binding entry in H-DMA.
Fig. 2 Attachment Procedure.
Fig. 3 (a) Intra-domain Location Update Procedure.

Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256 2253
Fig. 3 (b) Inter-domain Location Update Procedure.
C. Paging Procedure
Paging is the procedure though which CN finds the current IP address of MN. The signaling
procedures of paging when MN is in home network and foreign network can be described in Fig. 4 (a)
and Fig. 4 (b).
The paging procedure is initiated by CN with MN’s FQDN and can be divided into two stages. The
first stage is query and response in DNS system to get MN’s HoA. The second stage is getting MN’s
current IP address from the distributed mobility management related information in DMAs, as well as
tunneling packets to MN.
Fig. 4 (a) Paging when MN in home network
.
Fig. 4 (b) Paging when MN in foreign network

2254 Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256
D. Handoff Procedure
Handoff procedure is responsible for maintaining the communication continuity when the attachment
point of MN changes in communication. The signaling procedure of handoff can be described in Fig. 5.
When MN moving to another AR, previous AR detects the movement of MN and caches those
packets sent to MN before receiving Location Cancel message containing new IP address of MN. MN
reattaches to the Internet at new AR, initiating location update process by sending Location Update
Request to NAR, defined in B. Location Update Procedure. Upon receiving the new care-of address of
MN, PAR sends a Binding Update message to CN so that CN can send data packets to MN’s new IP
address. When PAR receives Binding Update Ack from CN, it will forward cached packets to MN’s new
IP address, and these packets will routing to MN’s NAR. At the end of location update process, presented
in B. Location Update Procedure., new AR sends a Location Update Accept message to MN indicating
the success of location update, and then NAR will forward packets to MN.
4. Simulations and Results
The simulation is completed in Open Chord and OverlaySim environment. Assume there are N nodes
in simulation, amongst these N nodes, when initiates the simulation, there are nnodes having capacity and
other N - n nodes only having load. Those n nodes can be considered as the servers distributed spread the
network with responsibility of mobility management. The number of neighbours involved in calculating
combined load and utility is denoted by k. For example, if k = 0, it means that no neighbour node would
join in, and if k = 1, it means that both the successor and predecessor of this certain node would be
combined to measure their load and utility in total. Therefore, the number of nodes nearby would be
combined equals to 2k,k for successor and k for predecessor.
The distributed mobility management mechanism proposed in this paper aims at overcoming the
disadvantages of the traditional centralized scheme. Hereby, the simulated properties include utility and
percentile.
Fig. 5 Handoff Procedure
Utility for each node is calculated as follows:
/utility load capacity (1)

Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256 2255
Percentile, a statistics term, is the value of a variable below which a certain percentage of
observations fall. It is defined as follows:
100
number of scores at X
number of scores below X
2
percentile
number of scores

<(2)
For the simulation parameters N=3000, n=300, and k = 0, the load and its percentile chart for load are
showed in Fig. 6. And the utility is given in Fig. 8(a).
Fig. 6 shows that all the loads are below 700, and seems that loads are evenly distributed to each
node. From the percentile chart in Fig. 6, at least 90 percent of loads are below 300, while only several
nodes have load over 350.
For the simulation parameters N=3000, n=300, and k = 4, the load and its percentile chart for load are
showed in Fig. 7. And the utility is given in Fig. 8(b).
From load chart in Fig. 7, it shows that all the loads (combined with 8 neighbouring nodes) are below
3000, and it seems that loads are evenly distributed to each node much better than the case k=0 which
means without neighbours. The percentile chart in Figure 14 shows that 90 percent of loads are in the
range from 500 to 2000, while only several nodes have load large than 2500. The case including
neighbour nodes achieves load balance better than the one involving no neighbours.
Fig. 6 Load and percentile for N=3000, n=300, k=0
Fig. 7 Load and percentile for N=3000, n=300, k=4
(a) k=0 (b) k=4
Fig. 8 Utility comparison for (a) k= 0 and (b) k = 4

2256 Shi Yan et al. / Physics Procedia 25 ( 2012 ) 2249 – 2256
Fig. 8 compares the utility for the cases of N = 3000, n = 300, k = 0 and N = 3000, n = 300, k = 4. The
results show that single node operating individually seems much busier than those distributed nodes
cooperating together with their neighbours. For individually working node, the utility changes from slight
larger than 0 to about 0.6 and several ones reach approximately to 1, while for the distributed cooperating
nodes, the utility falls into the interval from about 0.1 to 0.4, and no utility larger than 0.4.
5. Conclusions
In this paper, a MIPv6 based distributed mobility management mechanism is proposed. The
traditional centralized HA are distributed across DMAs (Distributed Mobility Agent). The mobility
management related binding data is defined accordingly. Some major signaling procedures such as
attachment, location update, paging and handoff are also introduced. Simulation results show that the
proposed approach can overcome the scalability problem of the centralized scheme. In addition, UE
reachability is also considered in this paper through extension to DNS servers.
References
[1] Mobile Internet Report. By MorganStanley. 2009.12.
[2] Xu Feng, Yan Xueqiang. Study on Flat Architecture for Mobile Network. Telecommunications Science. pp. 43-49. 2010.7.
[3] Semin Sim, Seung-Jae Han, Joon-Sang Park, et al. Seamless IP mobility support for flat architecture mobile WiMAX
networks. IEEE Communications Magazines, 2009, 6(6): 142-148.
[4] D. Johnson, C. Perkins, J. Arkko. Mobility Support in IPv6. IETF RFC 3775. June 2004.
[5] D. Liu, Z. Cao, P. Seite, H. Chan. Distributed mobility management. Draft-liu-distributed-mobility-02, July 2010.
[6] Yi Zou; Chakrabarty, K.. Distributed Mobility Management for Target Tracking in Mobile Sensor Networks. IEEE
Transactions on Mobile Computing. Vol.6, Issue 8, pp.872-887. 2007.8.
[7] Zygmunt J. Haas, Ben Liang. Ad Hoc Mobility Management With Uniform Quorum Systems. IEEE Transactions on
Networking. Vol. 7, Issue 2, pp.228-240. 2002.8.
[8] Bertin, P.; Bonjour, S.; Bonnin, J.-M. Distributed or Centralized Mobility?. Proceedings of GLOBECOM 2009. Pp. 1-6.
2009.12.
[9] Fischer, M.; Andersen, F.-U.; Kopsel, A.; Schafer, G.; Schlager, M., A Distributed IP Mobility Approach for 3G SAE.
IEEE 19th International Symposium on Personal, Indoor and Mobile Radio Communications, 2008. PIMRC 2008.
[10] Bertin, P.; Servane Bonjour; Bonnin, J.-M., A Distributed Dynamic Mobility Management Scheme Designed for Flat IP
Architectures. New Technologies, Mobility and Security, 2008. NTMS '08.
[11] JING Na-na, CHENG Dong-nian, HU Yu-xiang, WANG Hao-xue. Design and Implementation of Indirect
Communication Architecture Based on HIP Protocol. Computer Engineering. Vol.35, No.5, March 2009, pp.87-90.
[12] Fayza Nada (2007), Performance Analysis of Mobile IPv4 and Mobile IPv6. The International Arab Journal of
Information Technology, Vol.4, No.2, April 2007, pp.153-160.
[13] Vasileios Pappas, Dan Massey, Andreas Terzis, Lixia Zhang .A Comparative study of the DNS design with DHT-based
alternatives, INFOCOM 2006.