Report ID
2005-03
Report Authors
A. Meka and A. K. Singh
Report Date
Abstract
Sensor networks are distributed systems constrained by power andmemory limitations that can span large geographic regions. Thesenetworks can be used to monitor hazardous leaks such asradioactive waste or chemical plumes. We consider the importantproblem of tracking and querying a plume over such networks. As aplume moves, nodes of the network store its spatio-temporalsummary. We present a distributed index structure \\emph{DIST}, andalgorithms for range queries on the history of the plume. DISTlocalizes information with respect to time and space, and scaleswith the size of the plume. We then present an analytical costmodel for our querying schemes based on the query location, querysize, and plume distribution. Using this model, our adaptivescheme chooses the optimal querying scheme based on its predictedcosts. Experimental results show that DIST outperforms alternativetechniques in query, update, and storage costs, and scales wellwith the number of plumes.
Document
2005-03.ps580.43 KB