| Summary: | 碩士 === 國立臺灣大學 === 電機工程學系研究所 === 86 === The problems of determining a minimum cost connected subnetwork that spans
a given subset of vertices is known as the Steiner problem in networks. Since
the Steiner problem in networks is NP-complete, it is practical to develop
heuristics which find trees whose costs are close to optimal. The application
of Steiner heuristics in networks is used to construct multicast trees. In this
thesis, we survey previous heuristics of Steiner problems in networks and
multicast algorithms and propose a novel heuristic algorithm based on the
concept of edge sharing for the Steiner problem in networks. As it turn out,
the Stiner heuristics with edge sharing outperform the traditional Steiner
heuristics without edge sharing in turns of the Steienr tree cost, as our
simulation suggests. The simulation results indicate that our method performs
particularly well when the size of the multicast group is relatively small
compared to the total number of nodes in the network. The proposed algorithm
can directly be used to construct multicast trees without constraints.
|
|---|
