Multicast on Ad Hoc Networks

1
Multicast onAd Hoc Networks
2
Some cast terminology

• Unicast send packet from source to one node
• Multicast send packet from source to many nodes
• Broadcast send packet from source to all nodes
• Geocast send packet from source to all nodes
  inside a region
• Anycast access nearest of any of receivers
  sharing the same service
• Gossiping all-to-all broadcasting every node
  has a message and the goal is to get all messages
  to all nodes

3
Multicast over MANET

• Multicast is an essential technology for many
  applications
• such as group video conferencing and video
  distribution
• and it results in bandwidth savings compared to
  multiple unicast sessions
• Due to mobility of wireless nodes, topology of
  ad-hoc networks changes frequently
• Thus the established multicast tree is likely to
  be broken during the period of the transmission,
  causing interruption, pause, or jerkiness in the
  quality of received video signal
• As such, tree diversity is an effective technique
  to reduce the effects caused by tree failure
• however, it also reduces connectivity level as
  compared to single tree scheme
• Specifically, if double tree scheme requires a
  significant increase in node density in order to
  keep a high connectivity level
• it may be too expensive to implement in practical
  situations

4
Base Types of the MANET Multicast

• Flooding
• Advantages
• Simple
• It needs the smallest overhead
• Disadvantages
• It induces great traffic
• Its scalability is poor
• Tree-Based Routing
• Advantages
• Minimal data traffic
• Disadvantages
• The maintenance/update of the tree needs huge
  control traffic
• It scalability is poor

5
Flooding I
6
Flooding II
7
Flooding III
8
Flooding IV
9
Flooding V
10
Flooding VI
11
Features of the Sophisticated Multicast Protocols

• They utilize certain properties of the MANET
  network
• Tree-based multicast
• Utilize the ad-hoc property of the MANET, since
  it contructs the tree in a receiver-driven manner
• Mesh-based multicast
• Utilize the variability of the topology
• Stateless multicast
• Utilize the soft-state maintenance
• Location-aided multicast
• Utilize the knowledge of the position
• Using GPS or base-station-based methods
• Gossip-based multicast
• Utilize the randomness in the communication and
  the mobility

12
Distance Vector Multicast Routing Protocol (DVMRP)

• Originally it was designed for wired IP
• Modifications
• Dynamical grafting/pruning
• Check for the packet duplication
• The DVMRP maintains source-based trees
• Operation
• First flooding, then
• pruning back

13
Ad-Hoc On-Demand Distance Vector (AODV) and
Multicast-AODV (MAODV)

• Tree-first
• (shared data tree)
• In each group there is one leader
• Which periodically sends hello messages
• Starting join
• on the Fig. -gt

14
MAODV join
15
MAODV link failure 1-2-3
16
MAODV Evaluation

• Advantages
• Integration of
• the unicast AODV and
• the multicast functionalities
• Disadvantages
• Poor packet delivery under mobility
• Group Leader gt SPOF (Single Point of Failure)
• Shared-Tree

17
Ad Hoc MRP Utilizing Increasing ID-Numbers
(AMRIS)

• Tree-first
• shared data-tree
• It constructs and maintains a shared delivery
  tree rooted at a special node
• The construction of the shared tree is initiated
  by a special node and
• all members maintain the tree
• It applies multicast session membership
  identifier (msm-id) and core node (Sid)
• The tree is created after broadcasting the
  NEWSESSION messages
• The group members dynamically join to a multicast
  session
• The receivers know the identifier of each sender
• But this is not true vice-versa

18
AMRIS - join
19
AMRIS link failure 1-2-3
20
AMRIS evaluation

• Advantages
• No loops
• Link breaks are locally repaired
• Simplicity
• Disadvantages
• Waste of bandwidth
• Slow rejoin scheme
• Increased average hop distance

21
Core-Assisted Mesh ProtocolCAMP

• Mesh-first
• A mesh is constructed
• which has redundant path among the nodes,
  including possible loops
• It calculates reverse shortest path
• Each mesh-member periodically sends messages to
  the source on the shortest path
• These messages initiate joining to the mesh of
  that non-mesh-member MANET devices (nodes), which
  are on the path of the messages
• Some core nodes are responsible for accepting the
  join requests
• The nodes can join dynamically to the multicast
  group and they help to maintain the mesh
• The group members use a packet cache
• Using this packet cache they check
• their connectivity and
• rebuild their connections
• The receivers know the IDs of the senders
• But the senders do not know the IDs of the
  receivers

22
CAMP Operation

• One of more core nodes are designated for
  supporting the join procedure
• Here there is no need flooding
• oppositely to DVMRP
• The node check before joining to the multicast
  group if any of its neighbors is member of the
  multicast group
• If yes, the node announces its group membership
  request to its neighbors
• If not, the node sends a join messages to one of
  the core nodes
• The core nodes are not obviously needed
• If there is not available core node, then
• the node that wants to join applies an Expanded
  Ring Search (ERS)
• in order to find a node that is already member of
  the mesh

23
On-demand Multicast Routing ProtocolODMRP

• Mesh-first
• The sender initiates the construction of the mesh
• It is more robust than the MAODV
• But it has less scalability
• A source in each group sends periodically a
  join-query
• Using forwarding group
• Its packet delivery is better
• then the tree-based protocols
• However, maintenance of the mesh means overhead

24
Comparison of ad-hoc multicast routing protocols
25
Qualitative Comparison of Qualitative Comparison
of MAODV, ODMRP and Flooding
26
Gossip-based Protocols

• Some multicasting protocols use gossip as a form
  of probabilistically controlled flooding to solve
  several problems
• including network news dissemination
• The basic idea of applying gossip to multicasting
  involves having each member node periodically
  talk to a random subset of other members
• After each round of talk, the gossipers can
  recover their missed multicast packets from each
  other
• Similar to an epidemic trying to infect
  others with information
• To spread messages
• Or to recover missed messages

27
Gossip-based multicasting

• In contrast to deterministic approaches, a
  probabilistic scheme will better survive a highly
  dynamic ad hoc network
• because it functions independently of network
  topology and its nondeterministic property
  matches the networks characteristics
• Two main types
• Push style
• Each process talks periodically, to spread
  information it has received
• To a randomly chosen neighbour
• Pull style
• Each process talks periodically, to find out the
  new pieces of information
• To a randomly chosen neighbour

28
Anonymous Gossip

• The designers can apply the anonymous gossip
  multicast performance enhancement technique atop
  any tree-based or mesh-based protocol with
  minimal overhead
• This technique does not require a group member to
  have any knowledge of the other group members
• An anonymous gossip multicast protocol proceeds
  in two phases
• In the first phase, a protocol multicasts packets
  to the group
• In the second phase, periodic anonymous gossip
  takes place in the background
• as each group member recovers any lost data
  packet from other members of the group that might
  have received it