4. http://siti.ulusofona.pt 4
What is global mobility?
Today...
• we have widespread wireless networks, with different access
technologies
–heterogeneous networks, different network architectures
and protocols
• end users have highly nomadic lifestyle
• end users usually hold a cellular and a fixed access
subscription
• increasing demand for different services (from low-data-rate
non-real time to high-speed real-time applications)
• increasing variety of user-friendly and multimedia-enabled
terminals
7. http://siti.ulusofona.pt 7
Use-case 1: Hotspots/3G
A user may prefer to use a Wi-Fi connection when available
The system could choose which network is better when more
than one is available, based on user's preferences
The change of point of attachment should not require a re-
registration
9. http://siti.ulusofona.pt 9
Nomadism
• The same service look & feel delivered independently of user
location or terminal
• Change of attachment point requires service re-activation
• Sessions (e.g., TCP) have to be re-initiated
• Requires global form of addressing (identification)
and global AAA
• Example: 3G global roaming
Session continuity
• Sessions are seamessly diverted (handed over) from an old
identifier to a new
• Change of attachment point does not disrupt active sessions
• Examples: L2 mobility support (micro-mobility); Mobile IP
Main concepts
10. http://siti.ulusofona.pt 10
Location management:
• Locate a mobile user’s point of attachment for delivering
data packets
Handover management:
• Maintain a mobile user’s connection as it continues to
change its point of attachment
Main elements involved: Mobile Node (MN), Correspondent
Node (CN), Mobility Anchor Point (MAP)
Binding: notion of known-address and real address, and the
association between them
Main tasks
14. http://siti.ulusofona.pt 14
Network layer – Mobile IP
Home Agent (HA):
• Mobile IP aware router in the home network
• Mobility anchor point
• Control plane: signalling, registration, data redirection
• Data plane: routing
Mobile node (MN):
• MIPv6/v4 client
• If MIPv6, integrated in the IPv6 stack – not an additional
package!
Correspondent node (CN):
• Third-party with whom MN communicates
15. http://siti.ulusofona.pt 15
Mobile IPv4 (RFC 3344)
Agent discovery process: the MN determines if it is in the
home network.
Registration process: the MN tells its Home Agent it is away
and its current IP address (Care-of address) via Foreign
Agent.
Forwarding process:
the Home Agent
encapsulates all the
incoming packets
and sends them to
the MN's Foreign
Agent.
16. http://siti.ulusofona.pt 16
Mobile IPv6 (RFC 3775)
Binding Update (BU): association HoA - CoA
HA Registration: MN sends a BU to its HA
CN Registration: MN sends a BU to its CNs
Communication between MN and CN:
• Bidirectional tunneling
(through the HA): it uses
IP encapsulation
• Route optimization
(direct communication):
it uses Routing Header
type 2 extension.
17. http://siti.ulusofona.pt 17
MIPv6 – Advantages/Disadvantages
Advantages:
• Address auto-configuration: no need for a Foreign Agent
• Route optimization: binding updates (BUs) to CNs; avoid
triangle routing
• Routing header extension: less overhead than IP
encapsulation
• Ingress filter bypass: Home Address option
Disadvantages:
• Handoff latency: HA Registration and CN Registration
• Service disruption during handover
• Correspondent node Mobile IP aware
20. http://siti.ulusofona.pt 20
Issues in Mobility Management
Handover latency does not provide QoS required for real-time
applications
Host-based methods require protocol stack modification in
the mobile node (it increases the complexity, waste of
resources)
Network-based protocol needs more study and
standardization
A novel approach is needed to be integrated into both L2 and
L3 handover procedure
23. http://siti.ulusofona.pt 23
Last-hop towards the end user
Dynamic and spontaneous wireless environments
User owns and often carries devices that can be part of the
network
No clear splitting between network elements and end-devices
Human movement patterns are based on individual users
routines and movement target-interests.
Human behaviour may exhibit high variability (impacts the
network)
User-centric environments
25. http://siti.ulusofona.pt 25
Provide a new mobility management architecture more
suitable for user-centric environments
De-construct current mobility management based on
centralized and static mobility anchoring
Provide nomadic end-user experience
Provide session continuity
Main aspects:
• Handover optimization
• User-centric approach
Main aspects
26. http://siti.ulusofona.pt 26
Analyze and validate adequate cost functions and metrics to
estimate QoE levels on-the-fly
Intelligent selection of mobility anchor points:
•The best connection
•Users expectations
•Seamlessly, anytime, anywhere
•Characteristics (technology, interfaces, services)
•Environment (mobility behavior, social patterns, resources)
Handover optimization
27. http://siti.ulusofona.pt 27
Identify basic mobility functionalities on current standards
Decoupling of functionalities in different management
elements (mobility anchor points)
Evaluate the better location to place mobility management
elements (server, router, end device)
Management functionalities placed closer to the end-user
Optimize the distribution of mobility anchor points
Specify the mobility control mechanism required for
communication between the different entities
User-centric approach
28. http://siti.ulusofona.pt 28
Study of main standardized mobility management schemes
Study of user's behaviour in main user-centric scenarios:
Hotspot, MANETs, WMNs, UPNs, DTNs.
Identification of functional blocks
Categorization of functional blocks
Output: Deliverable 1 – UMM Use-cases
• Available on the website: http://siti.ulusofona.pt/~umm
• Section “Documents”
MM Characterization
29. http://siti.ulusofona.pt 29
Device Identification:
• MN's identification in the network
Identification database control:
• mechanism related to the recording of association between
known-address and real-address
Binding mechanism (registration/update):
• signaling to create/update a record (binding) in the
identification control
Routing/forwarding:
• process of intercepting packets destined to a known-
address, encapsulating and forwarding to real-address
Functional blocks
30. http://siti.ulusofona.pt 30
Handover negotiation
• process taken when the device has its real-address changed, in order
to guarantee that the user will keep active all its sessions during the
handoff process
Resource management:
• procedures to guarantee the quality of the connection when the MN
changes its point of attachment to the network
Mobility estimation:
• prediction of MN's movement, in order to prevent connection breaks
Security/privacy:
• security mechanisms to assure the integrity of the elements and
signaling of the mobility management system
Functional blocks
33. http://siti.ulusofona.pt 33
Approach 1: Control and data plane
• Control - signaling
• Data - data traffic, routing, forwarding and address
translation
Approach 2: Location and Handover management
• Location - locate devices, call delivery
• Handover - maintaining active sessions while MN is roaming
Decoupling functionalities
35. http://siti.ulusofona.pt 35
Data
Control
Located in the Home Agent
Functionalities:
Binding mechanism (registration and update)
ID database control
Handover negotiation
Security
Device identification
Routing/forwarding
Approach 1: Control and data planes - MIPv6
36. http://siti.ulusofona.pt 36
One approach: Siemens patent
Siemens – 2008
Idea: take the control functions off the home network and put
them into the premisses of a MSP (Mobility Service Provider)
Compatible with MIPv6 and MIPv4
Home Agent splitted into two elements:
• HA-Control plane (HA-C)
• HA-Data plane (HA-D)
Describes the communication between HA-C and HA-D
Approach 1: Patent Siemens
37. http://siti.ulusofona.pt 37
One approach: Siemens patent
Approach 1: Patent Siemens
HA-C: placed into the MSP premisses
Functionalities:
• MN registration
• Secure communication to MN
• Secure communication to HA-D
• Keep the mapping between HoAs and CoAs
• Decide which HA-D the MN should use
38. http://siti.ulusofona.pt 38
One approach: Siemens patent
Approach 1: Patent Siemens
HA-D: placed into the access router in the home network
Functionalities:
• Secure communication to HA-C
• Establishing a IP-in-IP tunnelling to the MN CoA
• Forwarding the packets to the MN CoA
40. http://siti.ulusofona.pt 40
One approach: Siemens patent
Approach 1: Patent Siemens
Advantages:
• There is no restriction into where to place the HA control
plane functionality
• Increased scalability
• Simpler processing on MIP-enable routers
Disadvantages:
• MN depends on the service provider (MSP) to offer the
mobility management service
• The overhead on the network for the signaling between
HA-C and HA-D is unknown
• The overhead on the time spent for the communication
between HA-C and HA-D is unknown
41. http://siti.ulusofona.pt 41
Why is this interesting?
Approach 1: Why splitting?
Splitting functionalities gives us the flexibility to choose
where to place each block of functionality
Splitting functionalities may divide the charge of work of the
involved elements
It is interesting to think about a decentralized approach
If it is possible to separate the mobility management into
control and data planes functionality blocks, maybe it is
possible to separate into other functionality blocks (for
instance, location and handover management)
42. http://siti.ulusofona.pt 42
Status of the project
Implementation of control/data splitting on NS-2
Definition of scenarios
1 poster published on international event
2 scientific papers published on international events
Next steps:
• Location/Handover management splitting
• Other functionality splitting approach, based on simulation
results
• Define positioning of splitted elements
44. http://siti.ulusofona.pt 44
Next generation wireless networks are focus of intense
research
User-centric environments are the part of the network where
the end user is located and deserve more attention
User-centric MM approach can make the system less
dependent on the access/service provider networks
Identified blocks represent basic functionalities to offer all the
specified requirements
There is no one approach that implements all functional
blocks
Conclusions
45. http://siti.ulusofona.pt 45
Handover optimization
• Influencing handover decision
• New metrics
• Consider user policies
Mobility patterns & models
• Mobility patterns influence architecture requirements
• Adequate support for de-centralized control requires a
better analysis of potential mobility patterns
Future research topics
46. http://siti.ulusofona.pt 46
Which OSI Layer(s) are best suited to support mobility
convergence?
Multi-access handover issues
• Resource management
• Dealing with simultaneous movement
• AAA, single sign-on
Future research topics