In on-board mobile networks, such as those proposed for (and employed in) public transport vehicles, users are connected to a local network that attaches to the Internet via a mobile router and a wireless link. Central and coordinated management of mobility in a single router, rather than by each user device individually, has numerous advantages; however, it also means that link outages, e.g. due to signal degradation or handoff failure, may have an immediate impact on a potentially large number of connections. We argue that the advance knowledge of public transport routes, and their repetitive nature, allows a certain degree of prediction of impending link outages, which can be used to offset their catastrophic impact. Focusing on the TCP protocol and its extension known as Freeze-TCP, we study how the performance of the protocol depends on the outage prediction probability. In particular, we propose a Markov model of Freeze-TCP and, using simulations, show that it accurately predicts the performance improvement gained by outage prediction.