This paper provides a QoS analysis of a dynamic, ubiquitous UMTS network scenario in the automotive context identified in the ongoing EC HIDENETS project. The scenario comprises different types of mobile users, applications, traffic conditions, and outage events reducing the available network resources. Adopting a compositional modeling approach based on Stochastic Activity Networks (SAN) formalism, we analyze the Quality of Service (QoS) both from the users' perspective and from the mobile operator's one. The classical QoS analysis is enhanced by taking into account the congestion both caused by the outage events and by the varying traffic conditions. The impact of users' mobility on the selected QoS indicators is further investigated combining the SAN modelling approach with an ad-hoc mobility simulator, which also allows to refine the model representing the UMTS network behavior.