Binary Information Press
The multiple-origins-multiple-destinations routing problem can be formulated as finding a minimal routes set satisfied with certain constraints in a given networks. Without constraints it can be simplified as a single-origin-multiple-destinations problem that finds a minimal Steiner tree for each source to its destinations. In the case of constraints the set of all minimal Steiner trees may unsatisfied the constraints, thus solving the problem become complicated. It is a typical NP-hard problem. This paper proposed a genetic algorithm with two evolving modules based on the idea of decomposing the complex problem into two objectives. One is to find multiple excellent Steiner trees for each source to its destinations. The other is to find an optimal combination of all Steiner trees.