The vision for sensor networks is to connect the physical world to existing systems using small, low-cost embedded, wireless devices. A fundamental problem in achieving this vision is developing a programming paradigm which is well-suited for programming sensor network applications. This problem is complicated by the severe resource constraints upon nodes and a diverse, evolving set of applications. Our research aims to provide an intuitive yet expressive programming paradigm for sensor networks that simplifies the process of reasoning about the multitude of nodes in the network by allowing programmers to reason about application behavior in terms of the physical entities in the environment.