A Multicast-based Protocol for IP Mobility Support

1
A Multicast-based Protocol for IP Mobility Support

• Ahmed Helmy, Assist. Prof.
• Electrical Engineering Dept
• Univ. of Southern California
• helmy_at_usc.edu
• http//ceng.usc.edu/helmy

2
Outline

• Problem Statement
• Mobile IP Overview and Shortcomings
• Multicast and IP-Mobility
• Paradigm Shift Multicast-based Mobility (MM)
• Protocol Mechanisms
• Performance Analysis
• Topology Models
• Movement Models
• Results
• Future Work

3
Problem Statement

• Providing efficient IP mobility support,
  especially for real-time applications
• Real-time applications are least tolerant to
  jitter and so are very sensitive to handoff
  delays
• Efficiency is measured in terms of
• handoff smoothness
• routing efficiency (end-to-end delays)
• network overhead (bandwidth consumed and links
  traversed)

4
Mobile IP
Foreign Agent (FA)
Foreign Network
Correspondent Node (CN)
Home Agent (HA)
Home Network

• When mobile node (MN) moves to a foreign network
  it obtains a
• care-of-address (COA) from the foreign agent (FA)
  that registers
• it with the home agent (HA)
• COA is used by HA to tunnel packets to MN

- Triangle Routing in Mobile IP - HA may be
needed to provides location hiding and
security - Inefficient in terms of network
overhead and end-to-end delays
5
Related work on Mobile IPv6
Foreign Agent 2 (FA2)
Correspondent Node
Foreign Agent 1 (FA1)
Home Agent (HA)
Home Network
Mobile Node

• With every move, the mobile node (MN) obtains a
  care-of-address
• (COA) and sends a binding update to the HA and
  the CN

6
MIP.v6 (router-assisted handoff)Previous
Location Approach
Foreign Agent 2 (FA2)
Correspondent Node
Foreign Agent 1 (FA1)
Home Agent (HA)
Home Network
Mobile Node

• With every move, the mobile node (MN) obtains a
  care-of-address
• (COA) and sends a binding update to the previous
  location/FA

7
Multicast and IP-Mobility

• Common issues in both paradigms
• Location independent communication/addressing
• Location discovery/management
• Packet forwarding

Location Independent Addressing

• IP-Multicast
• Single logical multicast group D-class address
• Senders do not know receivers
• Receivers do not know senders

• Mobile-IP
• Permanent home address
• Temp care-of-address(es)
• Address mapping done through the home agent

8
Location Management

• Mobile-IP
• Mobile node location
• Done thru home agent
• Meet thru registration of new address

• IP-Multicast
• Membership location
• Done thru IGMP routing
• Meet through the multicast tree

Packet Forwarding

• IP-Multicast
• Multicast forwarding
• Tunnel through the multicast tree (e.g., RP)

• Mobile-IP
• Unicast forwarding
• Tunnel through home agent

9
Paradigm Shift Multicast for Mobility

• Instead of obtaining a new COA and registering
  with the new foreign agent (and subsequently with
  the home agent) and de-registering the old
  address
• Use the same logical multicast group address and
  join/leave the group as you move

Potential Advantages

• Avoiding triangle routing problem
• Avoiding the need for home/foreign agents to
  continuously tunnel packets to the MN
• Smooth hand-off using standard join/prune
• Using shortest path (source-specific trees)

10
Multicast-based Mobility (MM) Architectural
Concept
Distribution tree dynamics while roaming
CN
CN Correspondent node (sender)
Wireless link
Mobile Node
11
Join/Prune dynamics to modify distribution
CN
CN Correspondent node (sender)
Wireless link
Mobile Node
12
Obtaining MNs multicast address

• A corresponding node (CN) obtains the multicast
  address of the MN through
• DNS lookup
• similar to getting the unicast (home) address of
  the MN
• requires update of DNS after allocation of
  multicast addresses to MNs
• Startup phase
• CN sends packets to home address
• Home agent encapsulates packets in multicast
  packets sent to the MN
• MN decapsulates these packets and sends a binding
  update to the CN with its multicast address

13
Startup scenario
Correspondent Node
Home Agent (HA)
Home Network
Mobile Node

• On first move, the mobile node (MN) sends a
  binding update to the CN

14
Main Protocol Mechanisms

• Mobile Node (MN)
• Join/Leave - Movement Detection
• Binding Updates - Care-of-address
• Decapsulation (during start-up phase)
• Base Station (BS) or first hop router
• Join/Leave - Caching and forwarding
• Sending beacons - Election (for robustness)
• Correspondent Node (CN)
• Binding update reception
• Home Agent (HA)
• Encapsulation (start-up) - Election (for
  robustness)

15
Performance Evaluation Route-based Analysis

• Performance metrics
• Network overhead
• End-to-end delay
• Handoff delay
• The model
• Topology model
• Movement model
• Multicast simulation (ns using centralized
  PIM-SM)

16
Topology Models

• Synthesized topologies
• random, transit-stub using GT-ITM and Tiers
  topology generators
• real topologies 2 Mbone, AS, ARPA maps
• 21 topologies with 47-5000 nodes, with various
  avg. degrees

17
Movement Models

• If MN is visiting node n, then node n1 is chosen
  according to one of the following movement
  patterns
• Random
• Neighbor next node to visit is randomly picked
  from nodes directly-connected to the currently
  visited node
• Cluster next node is randomly picked from one of
  6 nodes likely to fall within the same cluster as
  the current node
• For each movement pattern
• 100 movement steps in each simulation run, and 10
  runs with random selection of HA and CN

18
Performance Metrics
- Network overhead is proportional to total
number of links traversed - for Mobile IP ?(A
B), for MM ? C - we measure ?(A B) / ?
C for all simulation runs - End-to-end delay is
proportional to number of links traversed in each
simulation run - we define the ratio r
(AB)/C
19
Performance Metrics (contd.)
As the MN moves from node 1 to 2, the number of
added links L is 3 and the number of links to
previous location P (shown in dashed lines) is
2. As it moves from 2 to 3 there are no added
links (L0), and P is 2.
20
Overall Network Overhead
Total links traversed. ?(A B) / ? C 1.8
21
End-to-end Delay
Ratio r (AB)/C. Average r 2.11.
22
Handoff Latency

• MM proportional To L
• Mobile IP proportional B
• MIPv6 proportional C
• Previous location proportional P
• Define handoff latency ratios
• B/L, C/L and P/L

23
Handoff Latency for MM
Added links L. Average L 2.5 Links.
24
Handoff Latency Ratios
Average B/L, C/L and P/L ratios
25
Conclusion

• MM is quite simpler than Mobile IP (MIP)
  protocols. It re-uses many existing multicast
  mechanisms
• MM performs better than MIP and is more
  efficient (i.e. MIP was the beginning)
• Extensive simulations show that on average,
  compared to MIP
• MM incurs 1/2 network overhead
• MM incurs 1/2 end-to-end delay
• MM incurs less than 1/2 handoff delay

26
Issues and Future Work

• Multicast address allocation
• Security
• State overhead of the multicast tree
• Applicability requires ubiquitous multicast
• More detailed packet-level analysis (in progress)
• MM for intra-domain mobility?!
• Easier to manage/deploy
• Per-domain authentication
• Gains? Handoff performance!