Transcript DMAP - People

CS 6204 Paper Presentation
DMAP: A Scalable and Efficient
Integrated Mobility and Service
Management Scheme for Mobile
IPv6 Systems
Ing-Ray Chen, Weiping He, and Baoshan Gu
Paper Presented by: Vidhya Dass
10/10/2006
Agenda
Introduction
Contribution of the paper
DMAP model
Analytical Model
Numerical & Graphical Results
Applicability and conclusion
Introduction
MIPv6 : Network level protocol which is extension
of Mobile IP designed to authenticate MN using
IPv6 addresses
MN have permanent IP address on home network
MN roams into subnet acquire CoA (DHCP) from
that subnet
Binding update(address mapping CoA with MN’s
permanent IP) sent to HA(Special router on home
network)
CN -> HA(intercepted and tunneled) -> MN
Triangular routing avoided by MN sending binding
update to CN(address obtained from source header)
MN’s discovery of new subnet : Router supporting
neighbor discovery operational on each subnet.
Send router discovery message periodically.
MIPv6 Goal :
Enable mobility in IPv6
Maintain roaming connections in IP based
networks
Reduce overall network signaling cost
Approaches to reduce network signaling cost
MIP-Regional Registration (MIP-RR)
Hierarchical MIPv6 (HMIPv6)
Intra - Domain mobility management protocol
(IDMP)
MIP - RR
HA knows MN by Regional care of address (RCoA)
ie. GFA’s routable address
Local movement: MN’s CoA(Foreign agent address)
updated in Gateway Foreign Agent (GFA) , RCoA is
same
Regional movement : MN’s RCoA (new GFA IP
address) change informed to HA and CoA updated
in GFA
Drawbacks- Not consider service management
induced network cost
Visited Domain
FA
Home Network
GFA
HA
FA
MN
IP Network
HMIPv6
AR announce MAP(hierarchy of routers) identity by
means of router advertisement packets
Intra-regional Movement : CoA change propagated
to MAP, RCoA is same
MAP Domain boundary movement : RCoA change
propagated to HA and CN, CoA recorded in new
MAP
Drawbacks-MAP statically configured and shared by
all MN
IDMP
Domain
Region(HMIPv6,MIP-RR)
Mobility agent
MAP
Fast Handoff - MA multicasts packets to neighboring
agents during Handoff transient
Packets buffered at each SA(subnet agent) until
MN registers
paging support
MA initiates paging by multicasting solicitation
within the current paging area
packets buffered at MA until MN updates exact
location
Drawbacks
No mechanism to determine MAP domain size per MN
to reduce network signaling cost
Contribution of the Paper
Determine best DMAP domain size per MN dynamically
according to its mobility and service characteristics to reduce
network and signaling cost
DMAP
Extends HMIPv6
Dynamic Mobility Anchor Points(Access routers
chosen) for each MN
MN determines dynamically when and where to
launch DMAP for minimizing network cost
DMAP domain size depends on MN’s mobility and
service characteristics
HA and CN know MN by RCoA
Location Handoff : MN moves across subnet
boundary within DMAP region
Location + Service Handoff : MN moves across
DMAP boundary
Implement DMAP by DMAP table lookup design
using binding request messages defined in MIPv6
and HMIPv6
RCoA - CoA routing function performed by DMAP
through simple table lookup
Scaleable - All AR’s DMAP enabled
Assumption :
The AR of the first subnet that MN moves into after
DMAP domain change is chosen DMAP
After service area is crossed, if MN selects AR of
subnet just crossed as DMAP:
MN determines size of new service area
Obtains RCoA & CoA from current subnet
registers (RCoA,CoA) to current DMAP by binding
request message
 Inform HA and CN of new RCoA using standard
Mipv6
Packet delivery route: CN->DMAP->MN (tunneling
or direct)
MN’s service area - K, IP subnets
Goal : Dynamically determine optimal service area
(K) per MN
Special case : K is constant for all MN’s ???
- Degenerates to HMIPv6

: K is 1 ???
- Degenerates to MIPv6
DMAP : Integrated Mobility and Service Management in
MIPv6
Inter-Regional move(1 to 2):(Service+Location
Handoff)
AR of subnet B is new DMAP
MN’s service area - K subnets calculated
MN obtains RCoA and CoA from subnet B
Entry (RCoA , CoA) recorded in routing table of
AR of subnet B
HA and CN informed of RCoA address change
Intra-regional move(within 2) : (Location
Handoff)
MN acquires CoA from subnet
DMAP still in subnet B
DMAP informed of CoA address change
Tradeoff
Large Service area : DMAP not change often
Communication cost for service data delivery
high : CN->DMAP->MN
Location update cost is low
Small Service area : DMAP changed often
Communication cost for service data delivery low
Cost of informing HA and CN of DMAP change is
high
• MN lookup in built table :  and  as a function of
its location, time of the day and day of the week.
F(K) - Number of hops as a function of K(number of
subnets) : Determined dynamically by MN
Assumption :
Fluid flow model :
Average number of hops between 2
communicating models separated by K subnets is
Analytical Model
Find : Optimal service area using SPN
Why SPN
Deal with general time distribution of events
Deal with large number of states
Expressiveness to reason about MN’s behavior
Stochastic Petri Net Model
Intra-regional move
Register new CoA
with DMAP
MN obtains CoA
Move makes MN cross service area
Token in the place “moves” in SPN : Subnet
crossing event by MN
Mark(P) : Number of tokens in Place P
Mark(Xs) : Number of subnets crossed by MN since
it enters a new service area
 : One hop communication delay per packet in the
wired network
 : Ratio of communication delay in wireless to
wired network
F(Mark(Xs)+1) :Number of hops between current
subnet and DMAP( +1 for initial condition that
Mark(Xs)=0)
Transition rate of MN2DMAP
1/Communication time of MN informing DMAP of
new CoA
Communication
delay per packet in
the wireless network
CoA address change propagated to
DMAP in the wired network
Transition rate of NewDMAP
 : Average hop distance between MN and HA
 : Average hop distance between MN and CN
N : Number of CNs, MN concurrently engages
Communication time for MN to inform N CN’s
and HA in the wired network
Semi-Markov state representation( a , b )
a : Mark(Moves)
b : Mark(Xs)
Pi : Steady state probability that Mark(Xs) = i
where 1 i  K
C i,service : Network communication overhead to
service a data packet when MN in i th subnet in
service area
Communication
delay in the
wireless link
from the AR to
the MN
Delay between
the DMAP and a
CN in the fixed
network
Delay from
DMAP to the AR
of the MN’s
current subnet in
the fixed network
C i,location : Network signaling overhead to service
a location handoff when MN in i th subnet in service
area
Clocation : Average communication cost to service a
move operation by MN weighted by respective Pi
probabilities
i = K : Location +
Service to inform HA
and N CNs of RCoA
change
i < K : MN inform
DMAP of CoA
change
Total communication cost per time unit
 : Data packet rate between MN and CNs
 : MN’s mobility rate
Numerical Results
Basic MIPv6 - No DMAP
Communication cost for
servicing a packet delivery
Communication
delay in wireless
link from AR to
MN
Communication cost for
servicing a location handoff
Delay in the wireless
link from the MN to the
AR of the subnet that it
just enters into
Delay from that AR to
the HA
Communication delay
from CN to AR in
wired network
Delay from AR of the
subnet MN enters into,
to the CNs
Total cost per time unit for servicing data
delivery and mobility management operations
DMAP area large
(mobility cost
reduction)
DMAP degenerates
to HMIPv6
F(K) =
,  =  = 30 ,  = 10 and
normalized with =1
DMAP stays close
to MN to avoid CNDMAP-MN(service
cost reduction)
Degenerates to
Basic MIPv6
Kopt = Kh
DMAP degenerates to
basic MIPv6
* ,,CHMIPv6 CDMAPfor low SMR
(mobility management
cost dominates data
delivery cost)
* Threshold at which
DMAP degenerates to
HMIPv6
Conclusion:
DMAP incurs less
network overhead than
HMIPv6
Test Assumption
Average number of hops between DMAP and
MN separated by k subnets is F(k) =
Cost difference curves
are not sensitive to form
of F(k)
Assumption of F(k) =
justified
Applicability and conclusion
Novel DMAP for integrated mobility and
service management per MN
Procedure to find Kopt that minimizes overall
communication cost
MN dynamically looks up Kopt
DMAP outperforms basic MIPv6 at low SMR &
HMIPv6 at low and high SMR
Future : Test for sensitivity to other time
distributions
Thank you