Search in sensor networks: Challenges, techniques, and applications

Recently embedded wireless sensor networks (EWSN) have emerged as a solution to bridge the gap between the physical (and chemical and biological) world and the information world of the Internet. Deployment and effective use of EWSN posses numerous conceptual and technological challenges. In particular, there is a genuine need for the coordinated use of techniques from a variety of scientific and engineering fields, including DSP, computer science, information theory, communication, and applied mathematics. We address the problem of search in sensor networks using techniques from DSP and computer science. The search problem encompasses tasks such as indexing, stream searching, information filtering, semantic and syntax compression, profiling and recommending. The development of search tools will greatly facilitate efforts in many application of sensor networks, including event monitoring, low power operation, and physical space modelling. EWSN also poses unique challenges. For example, in the case of power and storage limitations, the key requirement is distributed space localized processing. The technical highlights of the paper include the first statistical stream search technique, the first technique for simultaneous search of correlated streams and the first technique for effective acquisition of correlated information under bandwidth and power constraints.