The potentially large population density and high handover rates of mobile users aboard a public transport vehicle underpin the need for robust and scalable quality-of-service (QoS) provisioning mechanisms designed for such environments. Traditional QoS approaches fall short of such criteria and have led to the proposal of a number of aggregation-based QoS mechanisms that maintain a single QoS state in the access network for the entire moving network subnet. Although such proposals can increase handover scalability, the dynamic connectivity and population of users within public transport vehicles can lead to bursts of requests being made due to passengers embarking on a vehicle in groups, intermittently increasing the control signaling density in the network and consequently increasing the cost to the operator. QoS aggregation policies alleviate this problem by buffering QoS requests at the mobile router installed on the vehicle for a period of time and then aggregating these into a single request to alter the resources reserved in the access network for the moving network. A number of such policies have previously been proposed in the literature to control and manage the buffering duration at the mobile router. However, since the aggregation decision of these policies is based on a static request-rate-dependent parameter, cost inefficiency can occur when the request rate is variable, as is typical of public transport vehicles due to passengers embarking and disembarking. This paper therefore provides a full mathematical and simulation-based analysis of a novel cost-driven aggregation policy and compares its performance with that of other policies previously proposed in the literature. Our results show that our cost-driven policy can reduce operator costs due to signaling by up to 21% compared with other policies while, at the same time, not putting the user at a disadvantage with long and unpredictable waiting times to establish a session.