For realistic simulations of ad hoc networks, the movements of the participating nodes must be considered. Many models have been proposed which generate movement patterns based on nodes' roles and activities. However, previous models mostly consider node activities individually for each node, making it difficult to precisely model joint activities. While advanced previous models are either based on the notion of schedules or on state machines, we propose a hybrid approach which combines the flexibility of the latter with the ability of the former to express time constraints. Further, as opposed to previous work, our model allows nodes to assume multiple roles and provides an intuitive way to reconcile their potentially conflicting schedules and state machines. Our model also considers the on/off behavior of nodes which is very important to evaluate adaptivity of protocols to node joins and failures under realistic conditions. A major contribution of our work is further that we take into account the fact that mobility, on/off patterns and traffic generation in a network cannot be modeled independently. Our comprehensive model is the first to include all three parts in one consistent form. Finally, it introduces the notion of home and preferred positions which allows very precise modeling of the general node distribution and positioning in an office scenario.