Information Technology Reference
In-Depth Information
Figure 12. Reviewed multicast protection algorithms.
4.1. Dynamic Multicast Protection Schemes
Two algorithms are suggested to reduce the session blocking probability and to decrease
the usage of network resources by removing or avoiding residual links in the topology that
consists of “light-tree” and its backup path.
4.1.1. Optimal Path-Pair Based Removing Residual Link (OPP-PRL) [32]
The OPP-PRL algorithm is derived from the Optimal Path-Pair based Shared Disjoint
Path (OPP-SDP). The OPP-SDP algorithm computes an optimal path pair between a source
and destination pair. In the OPP-PRL algorithm, a link-disjoint path pair is found for each
destination node. Using this algorithm, residual links between the source and destination can
be removed. In order to minimize additional costs and to increase sharing of new path pair
with those found previously, the OPP-SDP algorithm updates the cost to zero for the links
along the found path. Moreover, this algorithm decreases the usage of network resources by
removing the overlapping path and straddling path. However, paths are removed under two
conditions: (1) if there are two link disjoint paths, except the overlapping path. Moreover,
there is not any destination node in the overlapping path; and (2) there is not any destination
node in the straddling path. Results in [32] show that OPP-PRL can improve blocking
probability about 10% over OPP-SDP when traffic load is high.
4.1.2. Source-Leaf Path Based Avoiding Residual Link (SLP-ARL) [32]
In SLP-ARL, a primary tree is first created for a given multicast session, and then the
backup paths are found for the primary tree in order to avoid having residual links in the
procedure of finding backup paths for a given working tree. Therefore, we can say that there
are two stages in the SLP-ARL algorithm. In the first stage, a multicast routing algorithm is
used to create a primary tree in order to reduce the number of leaf nodes on the multicast tree.
Then, a backup path is derived for each destination node at the second stage. Hence, in this
algorithm, backup paths are only setup for the leaf nodes on the primary tree. Note that the
backup path and working path for a leaf node can overlap on some links, which have been
