Ad-hoc shared-ride trip planning is a planning task on a non-deterministic transportation network. We propose to solve this task in a mobile geosensor network, which consists of transportation clients and hosts. In a mobile geosensor network the communication costs are a critical factor. Trip planning agents need communication to collect knowledge about the current network, and any way to limit this need reduces the costs of a solution. This paper introduces a theoretical model based on time geography, where clients, as trip planning agents, can actively identify relevant transportation hosts before communication starts, and hosts can identify whether their route is relevant for a specific planning task before responding to any request. This model reduces the communication costs significantly, which is at first derived theoretically, and then confirmed by an example.