Deductive Approach for Programming Sensor Networks

As sensor networks become common and enable new applications, developing powerful paradigms for programming sensor networks becomes critical to realize their full potential as collaborative data processing engines. In this article, we propose a deductive framework for programming sensor networks, based on the observation that the sensor network can be viewed as a distributed database of facts gathered from the environment. We believe that the overall collaborative (distributed) functionality of a sensor network application can be easily represented using deductive (logic) rules, and the non-collaborative functionality can be embedded in built-in functions. The proposed framework is declarative, and more expressive than the previously proposed distributed database view of sensor networks. A fully-developed framework will allow the user specify with ease the high-level functionality of an application, while hiding from user the low-level details such as related to distributed computation, resource constraints, energy optimizations, etc. Our system translates a given high-level user program to energy-efficient distributed code that runs on individual network nodes. In this article, we motivate the deductive framework for programming sensor network applications, and develop query evaluation techniques for distributed evaluation of deductive queries in sensor networks. We present certain performance results that illustrate the robustness and efficiency of the generated distributed code.