PAQ: persistent adaptive query middleware for dynamic environments

  • Authors:

  • Vasanth Rajamani;Christine Julien;Jamie Payton;Gruia-Catalin Roman

  • Affiliations:

  • The University of Texas at Austin;The University of Texas at Austin;The University of North Carolina, Charlotte;Washington University in Saint Louis

  • Venue:
  • Proceedings of the 10th ACM/IFIP/USENIX International Conference on Middleware
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

Pervasive computing applications often entail continuous monitoring tasks, issuing persistent queries that return continuously updated views of the operational environment. We present PAQ, a middleware that supports applications' needs by approximating a persistent query as a sequence of one-time queries. PAQ introduces an integration strategy abstraction that allows composition of one-time query responses into streams representing sophisticated spatio-temporal phenomena of interest. A distinguishing feature of our middleware is the realization that the suitability of a persistent query's result is a function of the application's tolerance for accuracy weighed against the associated overhead costs. In PAQ, programmers can specify an inquiry strategy that dictates how information is gathered. Since network dynamics impact the suitability of a particular inquiry strategy, PAQ associates an introspection strategy with a persistent query, that evaluates the quality of the query's results. The result of introspection can trigger application-defined adaptation strategies that alter the nature of the query. PAQ's simple API makes developing adaptive querying systems easily realizable. We present the key abstractions, describe their implementations, and demonstrate the middleware's usefulness through application examples and evaluation.