栽钟右.PDF

Approximate Aggregation Techniques for Sensor Databases Jerey Considine, Feifei Li, George Kollios, and John Byers Computer Science Dept., Boston University fjconsidi, lifeifei, gkollios, byers g@cs.bu.edu Abstract In a typical sensor network, each sensor pro duces a stream of sensory observations across one or more In the emerging area of sensor-based systems, a sig- sensing mo dalities. But for many applications and sens- nicant chal lenge is to develop scalable, fault-toler ant ing mo dalities, such as rep orting temp erature readings, methods to extract useful information fr om the data the it is unnecessary for each sensor to rep ort its entire sensors col lect. An approach to this data management data stream in full delity. Moreover, in a resource- pr oblem is the use of sensor database systems, exempli- constrained sensor network environment, each message e d by TinyDB and Cougar, which al low users to p er- transmission is a signicant, energy-exp ending op er- form aggregation queries such as MIN, COUNT and ation. For this reason, and b ecause individual read- A VG on a sensor network. Due to p ower and range con-