An efficient evolutionary algorithm for (near-) optimal Steiner tree calculation: an approach to routing of multipoint connections

One of the most prevalent problems in modern telecommunication networks is the efficient routing of multicast connections. The formal formulation of this problem reduces to a graph-theoretical problem known as the Steiner tree problem, that is proved to be NP-complete. We present an efficient solution for the Steiner tree problem with application to multicast routing, based on a hybridized genetic algorithm with a hill-climbing technique that facilitates better local exploration of the search space. Moreover, an adaptive policy that is used to control the genetic algorithm parameters renders the proposed scheme efficient and adaptive under static and dynamic environment conditions. Using simulation, we came to the conclusion that our proposed algorithm compares favourably to some of the most effective deterministic approaches already known.