Location awareness is crucial for many mobile-network applications. While commercial applications rely heavily on the convenience and ubiquity of GPS, military applications must remain robust across the spectrum of denied and contested battlespaces. The use of interagent RF ranging measurements provides one means of reconstructing the relative network geometry. If all pairwise range measurements are always available to all agents, each agent can then separately solve for the network geometry. For dynamic mobile networks with constrained communications, and particularly for extended networks with many agents, the available range measurements may not uniquely specify the entire network geometry. Instead, each agent must discover and localize a solvable subset of the network. This report presents a decentralized method of rigid-neighborhood discovery and localization. The method is implemented in simulation under conditions of range-limited measurement and communication. Results suggest that rigid-neighborhood selection can improve relative localization compared to full-network or random-neighborhood selection.