Multicasting is an efficient information dissemination for a subscribed user group on networks. In this thesis the problem of multicasting in mesh-connected networks is studied. Having a group of processors distributed on mesh, the goal is to present a routing strategy such that every member of the group can effectively transmit data to the other members. The conventional strategies in solving the problem are source-based and shared-tree approaches, each having drawbacks in efficiency or scalability for large networks. To compromise between these two, we use the multi-shared trees approach. We apply a core-selection algorithm based on taxicab geometry to create a small number of distribution trees that are almost optimum with respect to multicast time and traffic.