You nailed it. The overwhelming majority of research around mesh networks assumes that nodes are stationary, and connection lifetimes are long-lived and consistent (if not perfectly reliable). My job is extra cool because none of those assumptions apply.