Sharma, Zadorozhny, Chrysanthis 1 Structural Health Monitoring With Whirlpool Divyasheel Sharma Vladimir I. Zadorozhny Panos K. Chrysanthis Info Sciences & Tele Dept. Info Sciences & Tele Dept. Computer Science Dept. University of Pittsburgh University of Pittsburgh University of Pittsburgh Pittsburgh, PA 15260 Pittsburgh, PA 15260 Pittsburgh, PA 15260 USA USA USA dsharma@sis.pitt.edu vladimir@sis.pitt.edu panos@cs.pitt.edu ABSTRACT We propose a novel data delivery strategy, called Whirlpool, for efficient SHM using Wireless Sensor Networks. Whirlpool implements a rotating interrogation of a monitoring structure and provides collision-aware scheduling of the monitoring queries. The Whirlpool strategy can be tuned for the required Quality of Data (QoD). We apply Whirlpool to examine the unique properties of output signals of the structure under critical integrity conditions and to perform instability detection using redundancy-based estimation of Kolmogorov-Sinai entropy. 1. INTRODUCTION SHM requires efficient collecting and analyzing of data obtained in response to ambient or forced excitation of the monitored system. Wireless sensor networks, which are easier to deploy than a wired sensor networks are a natural choice for implementing SHM [1]. However, SHM systems have special requirements with respect to efficient mechanisms for querying sensor data and delivering the query result in a timely manner. The data combined from all relevant sensors may be quite large and will require very high data transmission rates to satisfy time constraints. Meanwhile, limitations on sensor node resources like battery power imply that excessive transmissions in response to monitoring queries can lead to premature network death. An examination of the reasons that affect both energy consumption and response time in sensor monitoring queries reveals that (a) data transmission collisions represent a major source of energy and time waste in wireless communications; (b) unnecessary amounts of active time for the sensors, due to lack of synchronization among data transmissions, is another major source of wasted energy and time in wireless sensor networks. In order to address each of above issues we develop cross- layer query processing strategies that fuse techniques from different areas of databases and networking [2]. In this paper we propose a novel whirlpool data delivery technique, which tunes the sensor query processing for specific performance and quality of data requirements of SHM systems. Whirlpool query processing is based on splitting the sensor network into sectors and performing a rotating interrogation over the sectors such that the complete network is monitored. Whirlpool introduces a natural inter-sector concurrency when two or more sectors can be interrogated simultaneously. The query optimizer also schedules concurrent data delivery within each whirlpool sector (intra-sector concurrency). We demonstrate the high utility of whirlpool in non-intrusive SHM that observes the natural dynamics of structure for changes that indicate damage or instability [3, 4, 5]. In particular,