Skip to Main Content
Distributed actuation is a major application of sensor networks that relies on the information provided by the sensors and the ability of the actuators to change the environment, to try to achieve a set of desired conditions. In large scale ad hoc distributed sensor-actuator networks, traditional centralized control algorithms are undesirable due to limiting factors such as scaling, delays associated with collecting information, and energy consumption. To tackle distributed actuation problems in wireless sensor networks, groups of sensors and actuators can first be matched (clustered) efficiently and then the problem posed and solved in localized manners. In this paper, given a set of sensors and controllable sources of superposable phenomena (e.g. light and heat), we first formulate and optimally solve the distributed sensor-actuator problem as an instance of centralized quadratic programming. We then investigate localized heuristics that can achieve near optimal results while trading off message exchanges and energy consumption for accuracy and latency in obtaining the results. We also briefly discuss the clustering algorithm used as a heuristic for solving the matching problem associated with matching the sensors to the proper actuators. Light sensors and sources are used throughout the paper as an illustrating application.