skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: A Statistical Model and Computer program for Preliminary Calculations Related to the Scaling of Sensor Arrays

Abstract

Recent advances in sensor technology and engineering have made it possible to assemble many related sensors in a common array, often of small physical size. Sensor arrays may report an entire vector of measured values in each data collection cycle, typically one value per sensor per sampling time. The larger quantities of data provided by larger arrays certainly contain more information, however in some cases experience suggests that dramatic increases in array size do not always lead to corresponding improvements in the practical value of the data. The work leading to this report was motivated by the need to develop computational planning tools to approximate the relative effectiveness of arrays of different size (or scale) in a wide variety of contexts. The basis of the work is a statistical model of a generic sensor array. It includes features representing measurement error, both common to all sensors and independent from sensor to sensor, and the stochastic relationships between the quantities to be measured by the sensors. The model can be used to assess the effectiveness of hypothetical arrays in classifying objects or events from two classes. A computer program is presented for evaluating the misclassification rates which can be expected whenmore » arrays are calibrated using a given number of training samples, or the number of training samples required to attain a given level of classification accuracy. The program is also available via email from the first author for a limited time.« less

Authors:
Publication Date:
Research Org.:
Ames Lab., IA (US)
Sponsoring Org.:
USDOE Office of Nonproliferation and National Security (NN) (US)
OSTI Identifier:
797629
Report Number(s):
IS-5147
TRN: US200220%%1148
DOE Contract Number:  
W-7405/Eng-82
Resource Type:
Technical Report
Resource Relation:
Other Information: PBD: 1 Apr 2001
Country of Publication:
United States
Language:
English
Subject:
54 ENVIRONMENTAL SCIENCES; ACCURACY; CLASSIFICATION; COMPUTER CODES; DATA; PLANNING; SAMPLING; STATISTICAL MODELS; TRAINING; VECTORS

Citation Formats

Max Morris. A Statistical Model and Computer program for Preliminary Calculations Related to the Scaling of Sensor Arrays. United States: N. p., 2001. Web. doi:10.2172/797629.
Max Morris. A Statistical Model and Computer program for Preliminary Calculations Related to the Scaling of Sensor Arrays. United States. doi:10.2172/797629.
Max Morris. Sun . "A Statistical Model and Computer program for Preliminary Calculations Related to the Scaling of Sensor Arrays". United States. doi:10.2172/797629. https://www.osti.gov/servlets/purl/797629.
@article{osti_797629,
title = {A Statistical Model and Computer program for Preliminary Calculations Related to the Scaling of Sensor Arrays},
author = {Max Morris},
abstractNote = {Recent advances in sensor technology and engineering have made it possible to assemble many related sensors in a common array, often of small physical size. Sensor arrays may report an entire vector of measured values in each data collection cycle, typically one value per sensor per sampling time. The larger quantities of data provided by larger arrays certainly contain more information, however in some cases experience suggests that dramatic increases in array size do not always lead to corresponding improvements in the practical value of the data. The work leading to this report was motivated by the need to develop computational planning tools to approximate the relative effectiveness of arrays of different size (or scale) in a wide variety of contexts. The basis of the work is a statistical model of a generic sensor array. It includes features representing measurement error, both common to all sensors and independent from sensor to sensor, and the stochastic relationships between the quantities to be measured by the sensors. The model can be used to assess the effectiveness of hypothetical arrays in classifying objects or events from two classes. A computer program is presented for evaluating the misclassification rates which can be expected when arrays are calibrated using a given number of training samples, or the number of training samples required to attain a given level of classification accuracy. The program is also available via email from the first author for a limited time.},
doi = {10.2172/797629},
journal = {},
number = ,
volume = ,
place = {United States},
year = {2001},
month = {4}
}