According to [16] [30] [5], explicit multicast schemes are suitable for ad hoc networks,
where there are frequent changes in the network topology. These schemes differ from
traditional multicast schemes in that they are trying to reduce the amount of routing
information kept in the group members to perform multicast routing. A number of routing
protocols, discussed before, have been proposed based in this type of networks like DDM
[20], E2M [38] and EM2NET [7].
DDM [20] uses an extended header to include the list of destinations and their next hops.
Each next hop is attached to a list of corresponding destinations and these lists are
encoded into the Xcast header packet and broadcasted towards the source neighbours.
The nodes in the neighbourhood which receive this packet will process the header and
check if they are part of the next hop list. It is worth noticing that all the nodes in the
neighbourhood will process the received packet header. This issue will cause high
overhead in processing and delay in forwarding packets. In our proposed work, SARM,
only a set of nodes are involved in processing the packets and the remaining node just
forward the packet without any processing procedure. This feature will decrease the time
that the packets spend to be received by the destinations.
E2M [38] is proposed to improve the scalability issue in multicast ad hoc networks. It
employs the concept of Xcast Forwarder (XF) which is selected dynamically during the
message forwarding procedure. The XF selection depends on predefined threshold which
represents the number of destinations served by this XF. The major drawback in E2M is
that it does not solve effectively the node movement issue in ad hoc networks, especially
when XF moves which happens frequently in MANETs. In E2M, an XF moves in a way
that cause link failure, the source will be notified and all the destinations served by this
XF will be inserted into the next packet header created by the source node. This solution
is not efficient because the route is not already established for these destinations which
will cause a packet loss in the served destinations. In our proposed protocol in this thesis,
the node mobility issue is solved by performing local search for new connection, because
of link failure related to node movement. This solution will decrease the time for
99