Reconfiguration and self-adaptation are vital capabilities of networked embedded sensor systems that are required to operate in dynamic environments. Runtime technologies that allow software to evolve as system requirements change are critical to the development and deployment of such systems. This project aims at developing adaptive software by establishing a novel direction in software composition for networked embedded sensor systems based on Service-Oriented Architectures and Model-Integrating Computing. We are developing scalable solutions that enable reconfiguration in sensor networks based on distributed algorithms for constraint satisfaction problems, efficient reconfiguration architectures based on services. We are using an experimental sensor network test-bed consisting of heterogeneous nodes developed in the Model-Based Embedded Systems (ModES) Lab to demonstrate the research advances. The software is available in open-source format for non-commercial use.