Scalability and extended lifetime are two critical design goals of any large scale wireless sensor network. A two-tiered network model has been proposed recently for this purpose. This is a cluster-based network model composed of relay and sensor nodes. Relay nodes are placed in the playing field to serve as cluster heads and to form a connected network topology for information dissemination at the higher tier. The relay nodes are capable of aggregating data packets from the sensor nodes in their clusters and transmitting them to the sink node via wireless multi-hop paths. In this paper, we study the relay node placement problem in large scale wireless sensor networks. Our objective is to place the fewest number of relay nodes in the playing field of a sensor network such that (1) each sensor node can communicate with at least one relay node and (2) the network of relay nodes is connected. However, placement strategies realizing goals (1) and (2) do not provide any fault-tolerance as ...