up of multicast tree only at the start stage (exactly, the searching for first level BNRs from
the source), and BNRs will take the responsibility of the update and maintenance of the
remained part of the multicast tree. This means the operations of joining or leaving are
always implemented locally, i.e., being done between the destination triggering the join/
leave and its closest BNRs in existing multicast tree. Therefore, no remarkable join /leave
latency will be incurred in SReM.
Table 4.3 summarises the cost analysis of SReM, Xcast, Xcast+, SEM, HBH and REUNITE
schemes. SReM, SEM, HBH and REUNITE have control plane against the Xcast and
Xcast+, their cost of packet header processing is minimised. It is clear that SReM is the one
with the most advantages. Furthermore, compared to SEM, SReM has less control overhead
and lower join and leave latency. This will be confirmed by simulation in the following
section.
Table 4.3 Cost analysis of SReM, SEM, HBH, REUNITE, Xcast and Xcast+
Xcast |
Xcast+ |
SEM |
SReM |
HBH |
REUNITE | |
Multicast address allocation |
none |
medium |
medium |
medium |
medium |
none |
Multicast routing |
none |
low |
low |
low |
medium |
medium |
Control overhead |
none |
medium |
medium |
low |
medium |
medium |
Overhead by |
high |
low |
low |
low |
medium |
high |
Extra header |
^Hιgh |
medium |
low |
low |
low |
low |
Deployment |
Low |
low |
low |
low |
low |
low |
Join and leave |
high |
high |
high |
low |
medium |
medium |
81