Due to the fast-growing demand for battery-powered mobile devices, running computing-intensive programs on these devices could be more economical with the help of migrating processes to the idle devices in the vicinity. To migrate processes, user level implementation is an engaging policy in that it minimizes the dependency with the host system. However, resource management dependency is still inevitable even in user level implementation, and it becomes an important issue on how to resolve the legacy depen-dency problems after processes are migrated to idle devices. In this paper, we consider the processes with reserved sharing resources, and propose an approach to minimize the communication cost between the migrated processes and the resources they shared. A resource augmentation algorithm with approximation bound is proposed to decide the allocation of shared resources, and an algorithm with more restricted, yet plausible con-straints is proposed to decide an optimal allocation of shared resources in polynomial time. A series of experiments was conducted, and the results were very encouraging.