Transcript Adaptive NEMO Support Protocol

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 58, NO. 7, SEPTEMBER 2009
An Adaptive Network Mobility
Support Protocol
in Hierarchical Mobile IPv6 Networks
Sangheon Pack,
Member, IEEE,
Taekyoung Kwon, Yanghee Choi,
Senior Member, IEEE, and
Eun Kyoung Paik
Manuscript received February 21, 2006; revised July 23, 2007, April 3, 2008, August 23, 2008, and November 24, 2008.
First published February 18, 2009;
1
Outline
•
•
•
•
•
•
Introduction
Background
Adaptive NEMO Support Protocol
Performance Analysis
Numerical Result
Conclusion
2
Introduction
• In wireless/mobile networks, mobility management is a
key technology
– for maintaining ongoing sessions while moving and
– for locating the users’ current locations to correctly deliver data.
• Next-generation wireless/mobile networks will be based
on Internet Protocol (IP) technology
– different characteristics from existing cellular networks
• Network Mobility (NEMO) support
– the demand for Internet access while moving on vehicles is
constantly increasing
• trains, buses, and ships
– Supporting the mobile Internet services on vehicles
3
Introduction (cont’d)
• mobile network (NEMO)
– a network whose point of attachment to the Internet changes due to
mobility
– consists of mobile routers (MRs) and mobile network nodes (MNNs)
• The MR within a NEMO is in charge of mobility management.
• IETF NEMO Working Group
– standardized an IP layer mobility support protocol
called NEMO basic support (RFC 3963, 2005)
• reduce the binding update (BU) traffic at the expense of tunneling overhead
• inefficient packet routing
4
Introduction (cont’d)
• None of the proposal in the literature addresses the
problem of balancing the BU traffic and tunneling
overhead in an adaptive manner.
– in heterogeneous wireless networks, different types of MNs and
applications exist
• The proposed adaptive NEMO support protocol in this
paper is based on Hierarchical Mobile IPv6 (HMIPv6).
– employs adaptive BU procedures that
• jointly consider the mobility and session activity and
– reduce either BU traffic for a high mobility or
– packet tunneling overhead for a high session activity.
5
Background
• Mobile IPv6
– a host mobility support protocol
– each MN can be identified by its home address (HoA)
– When an MN moves to a foreign network, the MN can be
reached by its care-of address (CoA)
– Data packets that are destined to the MN are transparently
routed to its CoA
• By the MN’s Home Agent (HA)
– The CN can directly send data packets to the MN that has sent
Binding Update (BU) to it
• by maintaining the binding information between the MN’s HoA and the
MN’s CoA.
6
Background –
NEMO Basic Support protocol
• Mobile Network Node (MNN)
– Local fixed nodes (LFNs)
• a fixed node that belongs to a NEMO
• unable to change its point of attachment while maintaining ongoing sessions
– Visiting mobile nodes (VMNs)
• a mobile node (MN) or MR assigned to a home link
• does not belong to a NEMO and
• is able to change its point of attachment while maintaining ongoing sessions.
• mobile network prefix (MNP)
– an IPv6 prefix delegated to the MR by its home network, toward its
ingress link
– A MR informs its HA (HA_MR) of the care-of address (CoA) and MNP
• Establishing a bi-directional tunnel between the MR and its HA
– used by the HA_MR to intercept packets destined for an MNN within
the NEMO
7
8
9
10
Background –
Hierarchical Mobile IPv6
• Hierarchical Mobile IPv6 (HMIPv6)
–
–
–
–
for host mobility
enhances Mobile IPv6 by introducing the Mobility Anchor Point (MAP)
minimizes the amount of Mobile IPv6 signaling traffic outside the MAP domain
reduces the handoff latency
• an MN configures two CoAs:
– regional CoA (RCoA)
• an address on the MAP’s subnet.
• configured when a MN receives a router advertisement (RA) message, with the MAP
option containing MAP information.
• is not changed as long as the MN stays within the same MAP domain
– On-link CoA (LCoA)
• an on-link CoA assigned to the MN’s interface based on the prefix information advertised
by an access router (AR).
– If an MN hands off between ARs and changes its LCoA within the same MAP
domain, it needs to register the new LCoA only with the MAP.
• the RCoA BU with the HA and CNs makes the MN’s mobility transparent
to the CNs with which it is communicating.
11
12
Adaptive NEMO Support Protocol
1. Address Configuration
3. Binding Update to the Mobility Anchor Point (MAP)
2. Binding Update to the Home Agent
4. Packet Delivery Procedure
•
The color green is used in the following slides to mark the special design or usage in the
adaptive NEMO Support Protocol, which are different from other schemes in the
literatures.
13
Address Configuration
• The MR configures its LCoA and RCoA
– by means of an RA message with the MAP option.
• The MR relays the MAP option toward its subnet, which allows VMNs to
configure RCoAs.
• The VMNs’ LCoAs is configured based on the MR’s advertisement message that
containing the MNP
– If the VMN’s LCoA is registered at the VMN’s HA, packet delivery (PD) to the VMN follows
the NEMO basic support protocol.
• the MR/VMN register either its LCoA or RCoA with its HA depending on the
characteristics of the NEMO.
– Binding cache of HA_VMN
• HoA_VMN  LCoA_VMN or RCoA_VMN
– Binding cache of HA_MR
• HoA_MR  RCoA_MR, or LCoA_MR
• MNP  RCoA_MR, or LCoA_MR
• When the VMN registers its RCoA with its HA,
– the pinball routing problem can be eliminated.
– the RCoA BU induces more BU traffic than the LCoA BU.
14
(HoA_VMN, LCoA_VMN) or (HoA_VMN, RCoA_VMN)
HA_VMN
(HoA_MR, RCoA_MR) or
HA_MR (HoA_MR, LCoA_MR)
L
(RCoA_MR, LCoA_MR) and
may be (RCoA_VMN, LCoA_VMN)
15
Binding Update to the MAP
• the MR always performs the MAP BU procedure
– MAP binding cache: RCoA_MR  LCoA_MR
– needs to contain the MNP
• the VMN performs the MAP BU procedure only when the
RCoA_VMN BU to HA_VMN is triggered
– MAP binding cache: RCoA_VMN  LCoA_VMN (MNP::VMNEUI)
• routed to the HA_MR !!
• The MAP compares the destination address in the outer
header with the MNP
– before it sends the packets to the HA_MR.
• If the destination address is derived from the MNP, the MAP
adds an additional (tunnel) header where the destination
address is the LCoA_MR.
16
Binding Update to the Home Agent
and Packet Delivery Procedure
• not considering BUs (by MR) to CNs
– security concerns
– significant modifications to the CNs
• BUs with the MNP to CNs
– BU explosion problem
• session-to-mobility ratio (SMR)
– the ratio of the session arrival rate to the handoff rate.
– high the session activity is a more important factor
• aim to reduce the packet tunneling overhead rather than the BU traffic.
– Low  the handoff rate is dominant
• better to seek to reduce the BU traffic
17
SMR: low reduce BU traffic
high reduce PD overhead
18
SMR: low reduce BU traffic
high reduce PD overhead
SMR < δ1
SMR ≧ δ1
19
Performance Analysis
• Total cost (bytes*hops) when m LFNs and n VMNs:
– CT(m,n) = CBU(n) + CPD(m,n)
– since LFNs do not perform any BU registrations, the BU cost
does not consider the number of LFNs, m
• CBU(n|i, j) : the BU cost when a NEMO experiences i AR
subnet crossings and j MAP domain crossings during an
intersession arrival time.
• L: the number of packets of a session
20
• a : The per-hop transmission cost for a BU/BACK pair
• b : The per-hop transmission cost for a tunneling header
21
• NEMO Basic Support Protocol
22
• HMIPv6-Based NEMO Support Protocol
–
The same with the adaptive NEMO support protocol under the condition SMR ＜ δ1 and SMR ≧ δ2
23
25
• Adaptive NEMO Support Protocol
26
27
Pr(r< δ)
•
•
•
•
λI : The intersession arrival time follows an exponential distribution with rate λI
tI : the random variables for the intersession arrival time
tS be the random variables for the AR subnet residence time
SMR random variable r is given by tS/tI , and its mean is λI/μS. The SMR
cumulative distribution function is derived as [28]
( λS [28]= λI )
[28] S. Pack, X. Shen, J. Mark, and J. Pan, “Adaptive route optimization in hierarchical mobile IPv6 networks,”
IEEE Trans. Mobile Comput.,vol.6, no. 8, pp. 903–914, Aug. 2007.
28
α(i) and β(j)
• 1/μS and 1/μD : The AR subnet and MAP domain residence time of
a NEMO follows a general distribution with mean 1/μS and 1/μD
• fS*(s) is the Laplace transform of fS(t). (pdf. of the AR subnet residence time)
• Based on the fluid flow model [25]
–
–
–
–
v  2r v  2

2
  r
 r
v : the average velocity of a NEMO,
LS and LD : the perimeters of an AR subnet and MAP domain
AS and AD : the areas of an AR subnet and MAP domain
If we assume that a MAP domain consists of N subnets,
then μD = μS/ √N [27]
[25] X. Zhang, J. Castellanos, and A. Capbell, “P-MIP: Paging extensions for mobile IP,” ACM Mobile Netw. Appl., vol. 7, no. 2, pp. 127–141, Apr. 2002.
[27] K. Wang and J. Huey, “A cost effective distributed location management strategy for wireless networks,”
Wirel. Netw., vol. 5, no. 4, pp. 287–297, Jul. 1999.
[*] R. Thomas, H. Gilbert, and G. Mazziotto, “Influence of the movement of the mobile station on the performance of the radio cellular network,” in
Proc. 3rd Nordic Sem., Copenhagen, Denmark, Sept. 1988, paper 9.4.
29
• λI : The intersession arrival time follows an exponential distribution with rate λI
• Let NS and ND be the numbers of subnet crossings and domain crossings during
an intersession arrival time.
E[tc] = 1/λI [29]
• Pr(NS = i) = α(i) and Pr(ND = j) = β(j) are derived as follows [29]:
[29] Y. Lin, “Reducing location update cost in a PCS network,” IEEE/ACM Trans. Netw., vol. 5, no. 1, pp. 25–33, Feb. 1997. 30
Numerical Results
• The AR subnet (MAP domain) residence time is assumed
to follow a Gamma distribution with mean 1/μS (1/μD)
and variance VS (VD).
• the SMR cumulative distribution function
31
Numerical Results (cont’d)
• Optimal SMR Threshold
– find δ1∗ first because δ2∗ is dependent on it
– Let φ be the SMR value where CLCoAMR(r) = CRCoAMR(r)
– Then, the adaptive NEMO support protocol chooses the
RCoA_MR BU when the SMR is less than φ
32
large L
 the PD cost dominated
 high packet overhead
33
Numerical Results
– Effect of SMR
34
Numerical Results
– Effect of Number of VMNs (n)
BU&PD cost ↗
35
Numerical Results
– Effect of Session Length (L)
36
Conclusion
• an adaptive NEMO support Protocol based on HMIPv6 and NEMO basic
support protocols
– depending on the SMR
• a key factor characterizing a mobile network
– to reduce the number of BUs when the SMR is low
– the number of tunneling is reduced when the SMR is high
• Developed analytical models to evaluate NEMO support protocols and
• Verified the analytical results by simulations
– the proposed scheme
• achieves a significant performance improvement for a low SMR and
• shows comparable performance with the NEMO basic support protocol when the SMR is
high.
• As further research,
– to reduce implementation overhead due to SMR measurement and
– to address security issues
37
comments
• The loading of MAP increased
– + Comparing the destination of each packet to the MNPs
• Short life session, e.g., web browsing, but only consider
sessions initiated by the CN?
• MR deliver the current SMR to the VMNs?
– Calculating their own SMRs may be more implementable and
adaptable
• The system model developed in this paper can be used to
analyze macro+micro mobility
38