■ While working on this thesis, the scalable recursive multicast protocol (SReM),
has improved the scalability feature by introducing the branching node routers
(BNRs) to keep the routing state information instead of all group members. Also
this protocol reduced the join/leave implosion by presenting the locally and
dynamically join/leave process. These points have not been studied together and
have not been evaluated before.
■ In SReM, it can be observed that the average processing and delivering cost is
better than other protocols in the same area of study. This is because of the
reduction in the header size which made the cost for packet processing and
delivering small.
■ The recursive build of tree used in the proposed multicast protocols made the
join/leave cost small. The joining and leaving of group members is done locally
and dynamically which cause the average cost of this process the minimum
between the other related protocols.
■ The presented recursive multicast protocol is scalable with the number of
members in the multicast group (i.e., group size). This scalability can be
observed from the stability of the packet header size in the protocol with the
group size, and the reduction in the end to end delay and the normalized
overhead with the group size. The end to end delay of SReM compared to
Xcast+ is about 10%, moreover, the normalized routing overhead of SReM
compared to Xcast+ is about 8%. These quantitative comparisons give a
conclusion that SReM is a promising approach in the area of scalable multicast
routing protocols.
■ It was observed that the roaming process presented in the scalable recursive
multicast protocol guarantees the mobility feature over receivers. It can be
observed that roaming process will not affect the improved scalability feature in
the original protocol.
124