Efficient tracking of 2D objects with spatiotemporal properties in wireless sensor networks

摘要

Wireless sensor networks (WSN) are deployed to detect, monitor and track environmental phenomena such as toxic clouds or dense areas of air pollution in an urban environment. Most phenomena are often modeled as 2D objects (e.g., a fire region based on the temperature sensor readings). People model the objects by their properties, and like to know how the properties change over time. This paper presents a distributed algorithm, which uses deformable curves to track the spatiotemporal changes of 2D objects. In order to save the constrained resources in WSN, our distributed algorithm only allows neighboring nodes to exchange messages to maintain the curve structures. In addition, our algorithm can also support tracking of multiple objects. Based on the in-network tracking of deformable 2D curves, we show that many spatiotemporal properties can be extracted by the in-network aggregation. Our experimental results have confirmed that our approach is resource-efficient with regard to the in-network communication and on-board computation.