Querying of sensor data

Abstract

Advances in micro-electro-mechanical systems (MEMS) allow sensors, actuators, mini-processors and radio devices to be integrated on small and inexpensive devices, hereafter called sensor nodes. The deployment of a large number of sensor nodes in an area of interest presents unprecedented opportunities for continuous and untethered sensing in applications ranging from environmental monitoring to military surveillance to disaster relief. In this chapter, we present a database approach to tasking sensor nodes and collecting data from the sensor network. Similar to the Cougar [35] and TinyDB models [19], we view each sensor node as a mini data repository, and the sensor network as a database distributed across the sensor nodes. As in traditional database systems, users need not be aware of the physical storage organization in order to query data of interest; they should be able to collect sensor data by formulating declarative queries in a language similar to SQL. Figure 6.1 illustrates a widely used mechanism for tasking sensor networks. A user formulates a declarative query and sends it to the sensor network through a special-purpose node, referred to as the gateway. In the query dissemination phase, the query is forwarded wirelessly hop-by-hop from the gateway to the relevant sensor nodes. In the result collection phase, sensor nodes probe their sensor devices and propagate sensor readings back to the gateway. The query dissemination phase and the result collection phase must be carefully designed to account for severe communication and energy constraints in sensor networks. Wireless data transmission consumes orders of magnitude more energy than processing on a sensor node [29]. This suggests that battery-powered nodes could significantly increase their lifetime by exploiting their local processing capabilities to minimize the amount of data transmissions. In this chapter, we highlight the importance of in-network processing, i.e. evaluating a query inside the network, as opposed to collecting raw data and processing it centrally at the gateway. The rest of this chapter is organized as follows. Section 6.2 classifies queries into groups based on user-defined quality-of-service requirements. Section 6.3 presents techniques for propagating queries into a sensor network, whereas Sect. 6.4 focuses on two tightly coupled aspects of the data-collection phase, namely routing and innetwork processing. Section 6.5 overviews a data-centric approach to storing data within the network, and Sect. 6.6 presents concluding remarks. © 2007 Springer-Verlag Berlin Heidelberg.