Efficient detection of motion-trend predicates in wireless sensor networks

This work addresses the problem of efficient distributed detection of predicates capturing the motion trends of mobile objects evaluated with respect to a (boundary of a) polygonal region, in the settings in which the (location, time) data is obtained via tracking in Wireless Sensor Networks (WSN). Specifically, we discuss in-network distributed algorithms for detecting two motion-trend predicates: Continuously Moving Towards and Persistently Moving Towards: first for a single object, and then the corresponding variants for multiple objects. We also present methodologies which consider the energy vs. latency tradeoffs when multiple tracked objects are being considered for validating the monitored predicates. Our experiments demonstrate that our proposed technique yield substantial energy savings when compared to the na ve centralized and cluster-based approaches in which the raw (location, time) data is transthitted to a dedicated sink where the predicates are being evaluated. (C) 2016 Elsevier B.V. All rights reserved.