On optimal communication spanning trees in embedded Ethernet networks

In network design, the Optimal Communication Spanning Tree (OCST) problem is to find a spanning tree that connects the network entities and satisfies their communication requirements with minimal total cost. In an embedded Ethernet network, we organize the full duplex links into bundles and install these bundles into ducts. The traditional OCST problem does not take into account this fact. In this paper, we introduce a mathematical model of the communication cost of an embedded Ethernet network. We propose novel algorithms to find an OCST. Their principle idea is that they iteratively examine a set of neighboring trees that differ in one duct and choose the tree with the lowest cost. Finally, we evaluate the performance of the algorithms in terms of the quality of the solution found as well as the running time.