Download XIE02-slide

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
A Novel Distributed Dynamic Location
Management Scheme for Minimizing
Signaling Costs in Mobile IP
Authors: Jiang Xie, Ian F. Akyildiz
CS 6204
Oct 11, 2005
Presented by Hui Zhang
Agenda
► Introduction
► Distributed
& Dynamic Regional Location
Management
► Signaling Cost Function
► Optimal Regional Network Size
► Analysis Results
► Conclusion
Introduction
► Mobile
IP:
 Increasing Demands for mobile wireless access to
Internet applications.
 enables terminals to maintain all ongoing
communications while moving from one subnet to
another.
► Some
challenges:
 The signaling cost may become very expensive
►
Location updates may become very significant as the number
of MNs increases.
 The signaling delay may be very long
► If
the distance between the visited network and the home
network of the MN is large, the signaling delay for the location
registration is long.
Introduction (cont’d)
► Mobile
IP Regional Registration aims
 To reduce the number of signaling messages to
the home network
 To reduce the signaling delay when an MN
moves from one subnet to another
IETF: Internet Engineering Task Force
Care-of address
Care-of address
GFA: Gateway Foreign Agent
FA: Foreign Agent
MN: Mobile Node
HA: Home Agent
packets1
Packets2
(Tunnel)
Care-of address
GFA: Gateway Foreign Agent
MN: Mobile Node
FA: Foreign Agent
HA: Home Agent
CN :correspondent node
Disadvantages
► Centralized
System Architecture
 A centralized GFA manages all the traffic within a
regional network.
►
Mobile IP regional registration is more sensitive to the
failure of GFAs.
 The number of FAs under a GFA is very critical for
the system performance.
Distributed Dynamic Regional
Location Management
► Distributed




GFA management scheme
Each FA can function either as an FA or a GFA
The first FA should act as GFA
GFA maintains a visitor list
GFA relays all the home registration requests to the HA
► Dynamic
scheme
 There is no fixed regional network boundary for each
MN.
 The number of FAs under a GFA is not fixed, but
adjusted for each MN according to the user-variant and
time-variant user parameters.
 An MN decides when to perform a home location update
according to its changing mobility and packet arrival
pattern.
Architecture
Advantages
► The
traffic load is distributed
to each mobility agent
► The system robust is
enhanced
► Each MN has its own
optimized system
configuration FROM TIME
TO TIME
Protocol for MNs
►
►
►
Each MN keeps a buffer for
storing IP addresses of mobility
agents.
An MN records the address of
the GFA into its buffer when it
enters a new regional network
and then performs a home
registration
After the home registration, the
optimal number of FAs for a
regional network is computed
based on the up-to-date
parameters of the MN
Comparison
► ‘Distributed
system architecture’ and ‘dynamic
regional network’ are independent. So, there are
four possible combinations:
 Centralized system architecture and fixed regional
network,
 Distributed system architecture and fixed regional
network,
 Distributed system architecture and dynamic regional
network, and
 Centralized system architecture and dynamic regional
network
Comparison (cont’d)
Signaling Cost
Update Cost - CLU
 Home registration cost
 Regional registration cost
► Packet Delivery Cost – CPU
► Location
 The transmission cost
 Processing cost to route a tunneled packet from
the HA to the serving FA of an MN
Signaling Cost - CTOT
 CTOT = CLU + CPU
► Total
Home registration cost
Regional registration cost
Location Update Cost
► The
cost model of Centralized Scheme
► The
cost model of distributed fixed scheme
and distributed dynamic scheme
•M: m-th movement (the action an MN moves out of a subnet)
•Tf: the average time an MN stays in each subnet
•E[M]: the expectation of M, which is a function of k
•k: the number of FAs beneath a GFA, the size of a subnet
Packet Delivery Cost
Packet Delivery Cost (cont’d)
► The
total packet delivery cost per unit time
 Centralized Scheme
 distributed fixed scheme & distributed dynamic
scheme
Optimal Regional Network Size
optimal number of FAs beneath a GFA (kopt)
is defined as the value of k that minimizes the
cost function
► The iterative algorithm is used to calculate the
► The
kopt
 Because k can only be an
integer, the cost function is
not a continuous function of
k. Therefore, it is not
appropriate to take derivative
with respect to k of the cost
function to get the minimum.
Analysis Results
► Centralized
vs
Fixed Scheme
Distributed Fixed Scheme
 even under non-optimal
regional network size, the
distributed scheme always
performs better than the
centralized IETF Mobile IP
regional registration scheme
 the distributed scheme with
optimal regional network size
can further improve the
performance.
Distributed Fixed Scheme vs
Distributed Dynamic Scheme
► the
signaling cost of the
distributed dynamic
scheme is less than that
of both the distributed
fixed scheme using fixed
optimal regional network
size and using user-variant
optimal size.
Distributed Fixed Scheme vs
Distributed Dynamic Scheme
► the
distributed fixed
system always pays
higher cost than the
distributed dynamic
system
► The
Distributed
Dynamic regional
location management
is the better choice
Conclusion
► Introduced
a Distributed Dynamic regional location
management for Mobile IP
 A distributed GFA system architecture where each FA
can function either as an FA or a GFA.
► may
allocate signaling burden more evenly.
 A dynamic scheme to dynamically optimize the regional
network size of each MN according to its current traffic
load and mobility
► Analytical
results demonstrated that the signaling
bandwidth is significantly reduced through the
distributed system architecture compared with the
IETF Mobile IP regional registration scheme.
Thanks
Question ?