Network Topology Control and Routing under Interface Constraints by Link Evaluation

In this paper we introduce a new approach for controlling the topology of networks under interface constraints, along with routing the given traffic demands. Our approach is applicable to the situations in which each network node can form links with several potential neighbors. We formulate the routing problem as a multi-commodity flow problem for maximizing the throughput. We extend the multi-commodity flow problem by introducing some non-linear constraints which represent the non-feasibility of having links which can violate interface constraints. The additional constraints lead to a kind of competition among the links which cannot co-exist. The links are chosen so as to maximize the network throughput. We provide a polynomial time algorithm by doing linear approximation of the non-linear constraints. We propose topology control algorithms which use the initial multi-commodity flow formulation of routing as well as algorithms which use the modified linear version of multi-commodity flow formulation. Our simulation results show the efficiency of our approach in choosing a proper set of links from the potential links that give a feasible and good network topology.