madden_aggregation - Supporting Aggregate Queries Over...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
Supporting Aggregate Queries Over Ad-Hoc Wireless Sensor Networks Samuel Madden, Robert Szewczyk, Michael J. Franklin and David Culler University of California, Berkeley madden, szewczyk, franklin, culler @cs.berkeley.edu Abstract We show how the database community’s notion of a generic query interface for data aggregation can be applied to ad-hoc networks of sensor devices. As has been noted in the sensor network literature, aggregation is important as a data-reduction tool; networking approaches, however, have focused on application specific solutions, whereas our in- network aggregation approach is driven by a general pur- pose, SQL-style interface that can execute queries over any type of sensor data while providing opportunities for sig- nificant optimization. We present a variety of techniques to improve the reliability and performance of our solution. We also show how grouped aggregates can be efficiently computed and offer a comparison to related systems and database projects. 1. Introduction Recent advances in computing technology have led to the production of a new class of computing device: the wireless, battery powered, smart sensor. Unlike traditional sensors deployed throughout buildings, labs, and equipment everywhere, these new sensors are not merely passive de- vices that modulate a voltage based on some environmental parameter: they are full fledged computers, capable of fil- tering, sharing, and combining sensor readings. At UC Berkeley, researchers have developed small sen- sor devices called motes , and an operating system, called TinyOS, that is especially suited to running on them. Motes are equipped with a radio, a processor, and a suite of sen- sors. TinyOS makes it possible to deploy ad-hoc networks of sensors that can locate each other and route data without any a priori knowledge of network topology. As various groups around the country have begun to de- ploy large networks of sensors, a need has arisen for tools to collect and query data from these networks. Of partic- ular interest are aggregates – operations which summarize current sensor values in some or all of a sensor network. For example, given a dense network of thousands of sensors querying temperature, users want to know temperature pat- terns in relatively large regions encompassing tens of sen- sors – individual sensor readings are of little value. Sensor networks are limited in external bandwidth, i.e. how much data they can deliver to an outside system. In many cases the externally available bandwidth is a small fraction of the aggregate internal bandwidth. Thus comput- ing aggregates in-network is also attractive from a network performance and longevity standpoint: extracting all data over all time from all sensors will consume large amounts of time and power as each individual sensor’s data is indepen- dently routed through the network. Previous studies have shown [6] that aggregation dramatically reduces the amount of data routed through the network, increasing throughput
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 08/25/2011 for the course EEL 5937 taught by Professor Staff during the Spring '08 term at University of Central Florida.

Page1 / 10

madden_aggregation - Supporting Aggregate Queries Over...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online