Abstract-Cyber-physical systems typically involve a large number of mobile autonomous devices that closely interact with each other and their environment. Standard design and development techniques from the embedded domain fail to accurately model the dynamics of such systems and, hence, there is an increasing need for new programming models and abstractions. Component-based design approaches are a promising solution to manage the complexity of large-scale dynamic systems. However, existing such approaches either do not accurately model transitory interactions between componentswhich are typical of cyber-physical systems -or do not provide guarantees for real-time behavior which is essential in many safety-critical applications. To overcome this problem, in this paper, we present a component-based design technique based on DEECo (Dependable Emergent Ensembles of Components). The DEECo framework allows modeling large-scale dynamic systems by a set of interacting components. In contrast to other component-based design approaches from the literature, DEECo provides mechanisms to describe transitory interactions between components. We introduce necessary extensions to the DEECo design flow and integrate it with real-time analysis techniques that allow reasoning about timing behavior at the component-description level. Finally, we illustrate the simplicity and usefulness of our approach on a case study consisting of an intelligent crossroad system.