Chang, Hung-Chih (2010-12). Sensing Applications of Fluctuations and Noise. Doctoral Dissertation. Thesis uri icon


  • Noise and time-dependent fluctuations are usually undesirable signals. However, they have many applications. This dissertation deals with two kinds of sensing applications of fluctuation and noise: soil bulk density assessment and bacterium sensing. The measurement of Vibration-Induced Conductivity Fluctuations (VICOF) provides information about the bulk density and other parameters of soils. Bulk density is the physical property of soils that is important to both the agriculture and construction industries. The traditional measurements of soil bulk density are often time-consuming, expensive or destructive. To determine the soil bulk density without the above drawbacks, the VICOF measurement scheme was proposed. The research of VICOF in this dissertation includes two parts: the initial phase of study and the new methods and their theory. In the initial phase of study, the simple experiments, theory, and simulations of VICOF were tested for relations between the soil bulk density, wetness, salinity, and the VICOF data. Then, new measurement arrangements and their theoretical models were proposed to improve the weaknesses of the initial approach (such as large scattering of data due to loose and heavy contacts) and to calculate the relationship between the measured signals and the electromechanical transport parameters of the soils. The bacterium sensing study in this dissertation was proposed to explore simple, practical, rapid, sensitive, specific, portable, and inexpensive ways to detect and recognize bacteria by Fluctuation-Enhanced Sensing (FES). One such potential way of bacterium sensing is to analyze their odor. The research of bacterium sensing also includes two parts: the initial phase of study and the new methods and their theory. The initial phase study was proposed to explore the possibility of detecting and identifying bacteria by sensing their odor via FES with commercial Taguchi sensors. Then the subsequently developed new methods and their theory provide a simple way to generate binary patterns with perfect reproducibility based on the spectral slopes in different frequency ranges at FES. This new type of signal processing and pattern recognition is implemented at the block diagram level using the building elements of analog circuitries and a few logic gates with total power consumption in the microWatts range.

publication date

  • December 2010