arXiv2024
Distributing multipartite entanglement over a quantum network means routing it through a shared resource state. Existing measurement-based schemes search for a fresh path and re-verify the topology before every request, placing a network-wide classical exchange on the critical path of each one. We introduce DODAG-X, which removes it. A single destination-oriented directed acyclic graph spanning tree is computed once and reused across all requests, so each party's route is recovered by following parent pointers instead of by a new search. The per-request routing cost drops from $\mathcal{O}(N)$ to $\mathcal{O}(\sqrt{N})$ on symmetric grids and to $\mathcal{O}(\log N)$ on small-world networks for $N$ nodes, and only the $N-1$ tree links need be maintained under link loss. Routing on the sparse tree also shrinks the neighborhoods cleared to isolate the parties, lowering measurements per request by roughly 19\% on small-world graphs and up to 34\% on moderately dense, strongly rewired ones; on a fixed tree the two protocols use identical counts. We prove correctness for up to three parties with no restriction on topology, and prove a sufficient condition under which one application yields an $n$-party GHZ state for any $n$. We then delimit it, exhibiting requests outside the hypothesis whose output is multipartite entangled yet in a different local-Clifford class. Under a discrete-time Markov failure model the classical repair layer matches the reachability of full-graph re-search up to a failed-edge fraction of one half, and a coherence criterion relating tree depth to memory lifetime identifies the viable hardware platforms.