Efficient multicast for link-heterogeneous wireless mesh networks

Wireless mesh networks (WMN) have emerged as an economical means for delivering last-mile Internet access. Multicast is a fundamental service in WMNs because it efficiently distributes data among a group of nodes. Multicast algorithms in WMNs are designed to maximize system throughput and minimize delay. Previous work has unrealistically assumed that the underlying WMN is link-homogeneous. We consider one important form of link heterogeneity: different link loss ratios, or equivalently different ETX. We model different link loss ratios by defining a new graph theory problem, HW-SCDS, on an edge-weighted directed graph, where the edge weights model ETX, the reciprocal of link loss ratios. We minimize transmissions in a multicast by computing a minimum HW-SCDS in the edge-weighted graph. We prove HW-SCDS is NP-hard and devise a greedy algorithm for it. Simulations show that our algorithm significantly outperforms the current best WMN multicast algorithm by both increasing throughput and reducing delay.