Greedy, Perimeter-State Routing (GPSR), a routing algorithm for multi-hop wireless, mobile networks and sensor networks, addresses two challenges key to mobile routing: responsivity without unattractive routing protocol message complexity, and minimization of per-router state size.

Traditional distance-vector and link-state routing algorithms push transitive reachability information network-wide to reflect changes in link status. In the mobile environment, where links constantly appear and disappear as a consequence of limited radio range and router motion, this strategy generates prohibitive routing protocol traffic. GPSR is responsive under such constantly changing topologies without prohibitive message cost, because it uses *geography* rather than transitive reachability information.

Traditional routing algorithms' use of a per-destination routing table is not tractable on large, flat-routed, fully mobile networks, on which no address aggregation is possible--in such cases, each host must store state for each other host on the network. GPSR is nearly stateless, and stores progressively *less* state on progressively more densely connected networks, again because it routes on geography.

In this talk, I will frame the mobile routing problem, describe GPSR, present simulation results of GPSR, and put this work in context with respect to other mobile routing research.