A genetic algorithm-based multicast tree for routing in pub/sub system

In order to solve the routing problem in pub/sub system, a Genetic algorithm (GA)-based multicast tree approach, denoted by GAMT, is explored to build a Steiner multicast tree. GAMT uses GA and second-shortest paths as compensators to minimize the overall cost of the transmission and the time of routing construction. By using GA optimizer to find appropriate connections, GAMT can achieve a better approximation ratio of multicast tree than the algorithm proposed by Kou, Markowsky and Berman (KMB) without largely increasing time complexity. For testing the multicast tree algorithms, we use the method of Waxman to create thirteen random networks of different size. According to experimental results, GAMT not only achieved lower cost multicast tree than Average Distance Heuristic (ADH), KMB and the method of Melhorn in most cases, but also used a smaller computational time than ADH