A genetic algorithm for multicast mapping in publish-subscribe systems

In publish-subscribe systems, multicast is an efficient way to propagate information from the publishers to a group of subscribers. This paper studies the problem of mapping a large set of subscriptions into a fixed, smaller set of multicast groups in order to support efficiently the dissemination of events. Given the large search space, it is infeasible to obtain the optimal solution in reasonable time. To address this difficulty, the paper proposes and evaluates a genetic search solution for the mapping problem.