You are on page 1of 5

A Comparative Analysis on the Signaling Load of Proxy Mobile IPv6 and Hierarchical Mobile IPv6

Myung-Kyu Yi, Jin-Woo Choi, and Young-Kyu Yang


College of IT, Kyungwon University 5-3 Seromkwan, San 65, Bokjeong-Dong, Sujeong-Gu, Seongnam, Gyeonggi-Do, 461-701, South Korea {kainos,cjw49,ykyang}@kyungwon.ac.kr
AbstractIn this paper, we investigate the performance of the proxy mobile IPv6 and compare it with that of the hierarchical mobile IPv6. It is well known that performance of proxy mobile IPv6 is better than that of hierarchical mobile IPv6. For the more detailed performance analysis, we propose an analytic mobility model based on the random walk to take into account various mobility conditions. Based on the analytic models, we formulate the location management cost and handoff management cost. Then, we analyze the performance of the proxy mobile IPv6 and hierarchical mobile IPv6, respectively. The numerical results show that the proxy mobile IPv6 can has superior performance to hierarchical mobile IPv6 by reducing the latencies for location update and handoff.

I. I NTRODUCTION Recently, the rapid and widespread dissemination of powerful notebook computer and wireless communication promises to provide users with network access at any time and in any location. In order to communicate, all mobile devices must be congured with an IP address in accordance with the IP protocol and its addressing scheme. The problem occurs when a user roams away from its home network and is no longer reachable using normal IP routing. This results in the active sessions of the device being terminated. A natural solution is to use IP layer mobility. Mobile IPv6 (MIPv6)[1] is the IETF proposed standard solution for handling terminal mobility among IP subnets. It allows a computer to roam freely on the Internet while still maintaining the same IP address. Each Mobile Node (MN) is identied by two IP addresses: its Home Address (HoA) and its Care-of Address (CoA). The HoA is a static address that is used to identify higher layer connections. The CoA is a temporary IP address for routing purpose when the MN is attached to a foreign link. When an MN moves from its home link to a foreign link, it rst forms a CoA based on the prex of the foreign link. Then, the MN informs its Home Agent (HA) and any active Correspondent Node (CN) by sending a Binding Update (BU) message. The BU message contains the MNs HoA and its CoA. The HA needs to store this information in order to forward packets address to the MNs home address. Therefore, data packets addressed to the MN are routed to its home network, where the HA now intercepts and tunnels them to the CoA toward the MN. When the MN moves back to its home link, it will notify the HA to delete the binding. In MIPv6, however, the CoA of an MN changes whenever

it moves from one IP subnet to another. This could lead to frequent registrations with the HA. Thus, Hierarchical Mobile IPv6 (HMIPv6)[2] is proposed by IETF to reduce signaling cost. The HMIPv6 is a mobility management protocol aimed at reducing wireless signalling and improving handover performance while moving within a particular domain. It uses a new MIPv6 node called the Mobility Anchor Point (MAP) to handle Mobile IP registration locally. When an MN moves into a new MAP domain in HMIPv6 networks, it needs to congure two CoAs: an Regional Care-of Address (RCoA) on the MAPs link and an on-link CoA (LCoA). While the LCoA is simply referred to as the CoA in MIPv6, the RCoA is autocongured by the MN when receiving the MAP option. If the MN changes its current LCoA within a local MAP domain, it only needs to register the new address with the MAP. Therefore, the HA and external CNs need not be informed about local mobility within an IPv6 network. In HMIPv6, all packets addressed to the MNs RCoA are intercepted by the MAP and tunnelled to the MNs LCoA.
MN MAG AAA LMA CN

MN Attachment AAA Query AAA Reply Proxy Binding Update Proxy Binding Acknowledgement data packet data packet

Router Advertisement

Router Advertisement data packet

Fig. 1.

Message ow in PMIPv6

However, MIPv6 and HMIPv6 require additional stacks and signaling for the MN. This could bring overhead such as battery power and computation resource consumption. Network based mobility support mechanism is another approach to solve the problems and support the IP mobility. It is called Proxy Mobile IPv6 (PMIPv6)[3] and is based on MIPv6. PMIPv6 enables IP mobility for a host without requiring its participation in any mobility-related signaling. There are two main components for the PMIPv6, a Local Mobility Anchor

978-1-4244-2966-0/09/$25.00 2009 IEEE

(LMA) and a Mobile Access Gateway (MAG). The LMA performs home agent roles, as dened in MIPv6, for the MN in the proxy mobile IPv6 domain. The LMA is the topological anchor point for the MNs home network prex (HNP) and it maintains MNs binding state. The MAG performs mobility related signaling to the LMA for the MN and tracks the MNs movements. Fig. 1 illustrates the message ow of overall operations in PMIPv6. When an MN enters a PMIPv6 domain and attaches to an access link, the MAG retrieves the MNs prole using its current identier. Then, the MAG will send a Proxy Binding Update (PBU) message to the LMA in order to register the current point of attachment of the MN. Accordingly, a binding cache entry and a tunnel for the MNs home prex will be created. Then, the LMA reply Proxy Bind Acknowledgement (PBA) message with the MNs HNP. After receiving the Router Advertise (RA) message, the MN creates its IP address. For packet routing, the LMA will route all received packets over the established tunnel to the MAG. The MAG forwards these packets to the MN. Certainly, the MAG will relay all the received packets over the tunnel to the LMA and then they will be routed towards the CN. In this paper, we propose an analytic mobility model based on the random walk to take into account various mobility conditions for performance analysis of PMIPv6 and HMIPv6. Based on the analytic models, we formulate the location management cost and handoff management cost. Then, we analyze the performance of the proxy mobile IPv6 and hierarchical mobile IPv6, respectively. The rest of the paper is organized as follows. Section 2 introduces an analytic mobility model for performance evaluation and Section 3 formulates signaling cost functions using the analytic model. Section 4 shows the numerical results based on the analytic model. Finally, conclusions are presented in Section 5. II. A NALYTIC M OBILITY M ODEL In this section, we develop an analytic model to derive the cost functions and compare the performance of the PMIPv6 and HMIPv6 schemes. We assume the existence of a subnetbased cellular networks. Our network mobility model is represented by a bounded-degree, connected graph G = (V, E) where the node-set V represents the subnets and the edge-set E represents the access paths between pairs of subnet. Let N = |V | be the number of nodes or subents in the network G. For a node v V , let (v) denote the set of neighbors of v in G as shown in Fig. 2. We propose to use random walk on a connected graph G representing our network model. A random walk on a graph is stochastic process which occurs in a sequence of discrete steps. A random walk on the graph G induces a Markov chain MG as follows[4]. The states of MG are the nodes of G and for any two nodes i and j, the transition probability between the corresponding states is given by Pi,j . We denote (i) as the set of neighbors in subnet i. For all other subnets, Pi,i = 1 Pi,j = |(i)+1| for j (i), and zero otherwise.

Subnet 2 Subnet 1 Subnet 3

Subnet 4

Subnet 8

Subnet 7

Subnet 5 Subnet 6
Fig. 2. Wireless cellular network model

TABLE I T RANSITION PROBABILITY MATRIX AND STEADY- STATE PROBABILITIES (SSP) Subnets S1 S2 S3 S4 S5 S6 S7 S8 Transition Probability S3 S4 S5 S6 0.00 0.00 0.00 0.00 0.20 0.20 0.00 0.00 0.94 0.03 0.00 0.00 0.20 0.20 0.20 0.00 0.00 0.33 0.33 0.33 0.00 0.00 0.04 0.92 0.00 0.20 0.00 0.20 0.00 0.00 0.00 0.00 SSP 0.0369 0.0608 0.4046 0.0606 0.0361 0.3017 0.0611 0.0381

S1 0.33 0.20 0.00 0.00 0.00 0.00 0.00 0.33

S2 0.33 0.20 0.03 0.20 0.00 0.00 0.20 0.00

S7 0.00 0.20 0.00 0.20 0.00 0.04 0.20 0.33

S8 0.33 0.00 0.00 0.00 0.00 0.00 0.20 0.33

III. S IGNALING C OST F UNCTIONS In this section, we present analytic signaling cost function to compare PMIPv6 and HMIPv6. A. Analysis of Signaling Cost in HMIPv6

f HA d e Internet CN

MAP / LMA c AR / MAG

a AAA

Fig. 3.

Network Model

For simplicity, we assume that each LMA is located on the MAP as shown in Fig. 3. Moreover, we dene the distance parameters used for the performance evaluation as follows: a : The number of hops between MN and AR/MAG. b : The number of hops between AR/MAG and MAP/LMA. c : The number of hops between AR/MAG and AAA. d : The number of hops between MAP/LMA and HA. e : The number of hops between MAP/LMA and CN. f : The number of hops between HA and CN. h : The processing cost of binding update at the HA. m : The processing cost of binding update at the MAP. Pj,i : The transition probability from subnet i to subnet j. i : The steady state probability of state i estimates the location probability of a user in subnet i. Rr : The refresh rate that an MN renews its location. ti : The time duration that an MN stays in a subnet. We assume that performance metric is the total signaling cost which consist of the handoff management cost and location management cost in HMIPv6 as follows:
HM HM HM CT IP = CH IP + CL IP

B. Analysis of Signaling Cost in PMIPv6 We assume that performance metric is the total signaling cost which consist of the handoff management cost and location management cost in PMIPv6 as follows:
P P P CT M IP = CH M IP + CL M IP

(5)

We use Cl to present the location management cost in local binding update in PMIPv6. Based on Eq.(2), the average total signaling cost of PMIPv6 scheme as follows:
N P CL M IP N

=
i=1

i Cg +
j=1,j=i

j Cl Rr ti (6)

The rst term is the registration costs from the LMA to its HA. The second term is the average registration costs for an MNs movement from MAG i to MAG j in Eq.(6). Based on the message ow in PMIPv6, Eq.(6) can be rewritten as follows:
N P CL M IP

=
i=1 N

i ((4a + 4b + 2d)U + h + m ) j (2b U + m ) Rr ti


j=1,j=i

(1)

(7)

We use Cg and Cl to present the location management cost in global binding update and local binding update, respectively. Based on the random walk model, we can get the total signaling cost of HMIPv6 scheme as follows:
N HM CL IP N

Since the PMIPv6 employs the per-MN-prex model, the movement detection delay and address conguration delay are not incurred in PMIPv6. However, it requires authentication procedure before the MAG sends a PBU message to the LMA. Therefore, we can get the handoff management cost in PMIPv6 as follows:
N P CH M IP = i=1 j=1,j=i N

=
i=1

i Cg +
j=1,j=i

j Cl Rr ti (2)

j Pj,i (tRA + tAAA ) Rr ti

(8)

The rst term is the registration costs from the MAP to its HA. The second term is the average registration costs for an MNs movement from subnet i to subnet j in Eq.(2). For simplicity, we assume that the transmission cost for binding update message is proportional to the distance in terms of the number of hops between the source and destination mobility agents such as HA, MAP, CN and MN. Using the proportional constant U , Eq.(2) can be rewritten as follows:
N HM CL IP

Based on the message ow for authentication process in PMIPv6, Eq.(8) can be rewritten as follows:
N P CH M IP N

=
i=1 j=1,j=i

j Pj,i ((3a + 2c)U ) Rr ti IV. N UMERICAL R ESULTS

(9)

=
i=1 N

i ((4a + 4b + 2d)U + h + m ) j (2(a + d)U + m ) Rr ti


j=1,j=i

In this section, we will demonstrate some numerical results. Table I and II show some of the parameters used in our performance analysis that are discussed in [4], [5], [6], [7], [8].
TABLE II P ERFORMANCE ANALYSIS PARAMETERS Parameter ti h U b d tM D Value 0.01 - 1000 30 0.1 3 10 0.2 - 0.6 Parameter Rr m a c e, f tDAD Value 0.01-1000 20 1 - 10 1 - 10 10 1

(3)

In HMIPv6, IP handoff latency can be expressed as the sum of the movement detection delay and address conguration delay. Therefore, we can get the handoff management cost as follows:
N HM CH IP = i=1 j=1,j=i N

j Pj,i (tM D + tDAD ) Rr ti (4)

We dene the relative signaling cost of PMIPv6 scheme as the ratio of total signaling cost for PMIPv6 scheme to that

Relative Signaling Cost

of HMIPv6 scheme. A relative cost of 1 means that the costs under both schemes are exactly the same.
0.968 Relative Signaling Cost 0.967 0.966 0.965 0.964 0.963 0.962 0.961 0.960 1 2 a PMIPv6 HMIPv6

0.935 0.934 0.933 0.932 0.931 0.930 0.929 0.928 0.927 0.926 0.925 1 2 3 4 5 6 7 8 9 10 c PMIPv6 HMIPv6

Fig. 6.

Effect of c on the relative signaling cost

Fig. 4.

Effect of a on the relative signaling cost

Fig. 4 shows the impact of delay between the MN and AR/MAG, a, on the relative signaling cost for b = 3, c = 3, Rr = 0.1, and ti =10. We can see that the performance of the PMIPv6, on the whole, results in the lowest total signaling cost compared with HMIPv6. This is because wireless link delay for binding update is high, which incurs a high signaling cost in HMIPv6. However, PMIPv6 is least affected because the MN is not involved in mobility-related signaling.
0.965 Relative Signaling Cost 0.960 0.955 0.950 0.945 0.940 PMIPv6 0.935 0.930 0.925 10 -2 Rr -1 10 HMIPv6

ti = 10. We can see that PMIPv6 results in the lowest total cost compared with HMIPv6 scheme. From the above analysis of the results, the PMIPv6 scheme has a considerable performance advantages over HMIPv6 scheme. So, we conclude that the PMIPv6 achieves signicant performance improvements by reducing the IP latencies of location update and handoff. V. C ONCLUSIONS In this paper, we studied performance of PMIPv6 and compare it with that of HMIPv6. For the more detailed performance analysis, we proposed an analytic mobility model based on the random walk to take into account various mobility conditions. Based on the analytic models, we formulated the location management cost and handoff management cost. Then, we analyzed the performance of the proxy mobile IPv6 and hierarchical mobile IPv6, respectively. The numerical results show that PMIPv6 can has superior performance to HMIPv6 by reducing the IP latencies for location update and handoff. ACKNOWLEDGMENT This research was supported by the Ministry of Knowledge Economy, Korea, under the ITRC(Information Technology Research Center) support program supervised by the IITA(Institute of Information Technology Advancement) (IITA-2008-C1090-0801-0040) R EFERENCES
[1] D. B. Johnson, C. E. Perkins, and J. Arkko,Mobility support in IPv6, IETF Request for Comments 3775, June 2004. [2] H. Soliman, C. Castelluccia, K. El-Malki, and L. Bellier, Hierarchical Mobile IPv6 Mobility Management (HMIPv6), IETF Request for Comments 4140, August 2005. [3] Sri Gundavelli, Kent Leung, Vijay Devarapalli, Kuntal Chowdhury,Proxy Mobile IPv6, IETF Request for Comments 5213, August 2008. [4] Taehwan Choi, Laeyoung Kim, Jeongeun Nah, and Jooseok Song, Combinatorial Mobile IP: A New Efcient Mobility Management Using Minimized Paging and Local Registration in Mobile IP Environments,Wireless Networks, Vol. 10, Issue 3, pp. 311-321, May 2004. [5] Wu, Z.D, An Approach for Optimizing Binding Lifetime with Mobile IPv6, in Proceeding of 28th Annual IEEE International Conference on Local Computer Networks, pp. 82-88, Oct. 2003.

Fig. 5.

Effect of Rr on the relative signaling cost

Fig. 5 shows the Rr , on the relative signaling cost for a = 1, b = 3, c = 3, and ti = 10. As shown in Fig. 5, the relative signaling cost increases as Rr increases. We can see that the performance of the PMIPv6, on the whole, results in the lowest total signaling cost compared with HMIPv6 scheme. These results are expected because the PMIPv6 scheme tries to reduce the signaling loads by reducing the address conguration delay. Moreover, the PMIPv6 does not require tunnelling overhead over the air. Fig. 6 shows the impact of delay between the MN and AAA, c, on the total signaling cost for a = 1, b = 3, Rr = 0.1,

[6] Jiang Xie, Akyildiz, I.F, A Novel Distributed Dynamic Location Management Scheme for Minimizing Signaling Costs in Mobile IP, IEEE Transactions on Mobile Computing, pp 163-175, 2002. [7] Akyildiz. I.F, Yi-Bing Lin, Wei-Ru Lai, Rong-Jaye Chen, A New Random Walk Model for PCS Networks, IEEE Journal on Selected Areas in Communications , vol. 18, pp1254 - 1260, July 2000. [8] Sheldon M. Ross, Introduction to Probability Models : Eighth Edition, Academic Press, 2002.

You might also like