Adaptive sampling for multiscale environmental sensor networks

Wietsma, Tristan

Permalink

https://hdl.handle.net/2142/30898 Copy

Description

Title

Adaptive sampling for multiscale environmental sensor networks

Author(s)

Wietsma, Tristan

Issue Date

2012-05-22T00:13:56Z

Director of Research (if dissertation) or Advisor (if thesis)

Minsker, Barbara S.

Department of Study

Civil & Environmental Eng

Discipline

Environ Engr in Civil Engr

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Adaptive Sampling
• Environmental Sensor Networks
• Nyquist-Shannon Sampling Theorem
• Hot Moments

Abstract

Environmental sensor networks enable researchers to collect data at an impressive order of magnitude, both temporally and spatially. Without effective sampling logic, these powerful tools can produce an overwhelming quantity of data that may not capture the most valuable information for scientific discovery. To address this issue, this research expands the definition of a “hot moment”, a term previously used to describe times of high biogeochemical activity, to include periods of elevated signal complexity, which is when dense data collection is most needed. Under this new definition, an indicator for hot moment identification is developed. Using this indicator as a performance metric, a family of frequency-based adaptive sampling models are developed that operate at different network scales. These algorithms make use of the Nyquist-Shannon sampling theorem, a fundamental contribution from the field of signal processing, and take advantage of the resource (energy, bandwidth, computation) and information advantages specific to the local (sensor), regional (base station), and global (the Cloud, i.e. distributed computing clusters across the network) network scales. The models are tested over historical soil moisture data. Results indicate substantial advantages to adaptive sampling relative to traditional fixed-rate (uniform) sampling in both data reduction and improved sampling over hot moments.

Graduation Semester

2012-05

Permalink

http://hdl.handle.net/2142/30898

Copyright and License Information

Copyright 2012 Tristan Wietsma

Owning Collections