4.6.1.1 Simulation environment and results
In this simulation, SReM is compared with Xcast, Xcast+ and SEM where comparison with
HBH and REUNITE is excluded. This is because of the following reasons:
• Xcast, Xcast+, SEM, HBH and REUNITE is considered the most efficient and recent
protocols in the area of scalable multicast routing.
• Xcast, Xcast+, SEM and SReM share the same objective in providing scalable
multicast routing explicitly.
• HBH and REUNITE have not been chosen to compare to our proposed protocol
SReM. This is because in both HBH and REUNITE the join/leave messages are sent
periodically , which results in some unfairness and inconvenience to compare with
the Xcast, Xcast+, SEM and our proposed SReM.
In this simulation, the Waxman’s probability model [67] is used to produce random network
topology. The average connection degree at nodes is between 3 and 4. The link costs are
chosen randomly and uniformly distributed in [1, 10]. We assume that there is one multicast
source, which is randomly chosen from the routers, and some of nodes are randomly selected
as LMRs, where a router becomes a LMR as long as there is one multicast receiver attached
to it. The link costs between receivers and LMRs are set a fixed value of 0.1, based on the
consideration of the local link cost is usually lower than that between routers. The source and
routers (BNRs) which are taking part in multicasting are called as active multicast routers
(AMR). The total number of AMR is changing with the joining / leaving of the member of
multicast group.
In our simulation model, to deliver multicast packets, the Dijkstra algorithm [4] is used to
find the shortest paths between the source and the receivers. This algorithm is often used in
82