Quality-aware data abstraction layer for collaborative 2-tier sensor network applications

Abstract

This paper presents PRIDE, a novel data abstraction layer for collaborative 2-tier sensor network applications. PRIDE, more specifically, targets distributed real-time applications, in which multiple collaborative mobile devices have to analyze a global situation by collecting and managing data streams from massive underlying sensors. PRIDE at these devices hides the details of underlying sensors and provides transparent, timely, and robust access to global sensor data under highly dynamic and unpredictable environments of emerging sensor network applications. For transparent and efficient sharing of global sensor data, a model-based predictive replication mechanism is proposed and integrated into a conventional data management system that supports diverse types of spatial and temporal queries. In addition, for robust and timely query processing, the predictive replication scheme is extended to the problem of guaranteeing Quality-of-Service (QoS) by introducing feedback control of the accuracy bounds of models. We show the viability of the proposed solution by implementing and evaluating it on a 2-tier sensor network testbed, emulating collaborative search-and-rescue tasks with realistic workloads. Our evaluation results demonstrate that PRIDE can achieve timely sensor data sharing among a large number of devices in a highly robust and controlled manner.