Bringing real world applications for wireless sensor networks into the classroom: Telemetric monitoring of water quality in an artificial stream [2012]
Page:
14
The following text was automatically extracted from the image on this page using optical character recognition software:
In order to determine how accurate the readings for our sensor cluster actually are the correlation between the data points for each probe type was determined. We also calculated the percent error for our probes treating the YSI sondes as the standard to determine variation. pH DO Temperature Correlation 0.615358 0.807081 0.755728 % Error 4.28714 16.4226 3.93644 VII. Discussion We began by using the Arduino and the Xbee to build our sensor cluster. Once we connected all three probes to the our wireless sensor cluster, readings on the dissolved oxygen sensor became completely inaccurate and the pH probe began returning unreliable data. It looked like the problem could be solved using code. Yixin Gu built the first of our sensor cluster circuits. When we tested that sensor cluster in a working aquarium we found that the probes all collected data correctly when placed in the water one at a time but the same inconsistency arose when we had multiple probes in the water. At this point Yixin began exploring ways to isolate the pH sensor and dissolved oxygen sensor physically. The final solution involved splitting the DC power supply with a transformer and using a timing chip to control when the pH sensor and dissolved oxygen sensor were on. This seemed to resolve the issue that we were seeing. The implication here is that the issue was tied to the water carrying the current produced by the two sensors and creating interference leading to inaccurate readings. When we look at the data from the field, we can see that the graphs track very similar patterns for each of the probes. The readings for temperature are all very similar from each of the three probe with a correlation of 0.755728 and a percent error just under 4. That is not surprising because it is the simplest circuit and provides the most straightforward data. Even at that, though, you see some variance. Those differences will be magnified in the pH and dissolved oxygen readings because each of those are temperature dependent. The pH readings are fairly accurate with only 4 percent error, but they fluctuate a good bit and have a relatively low correlation. The values returned by the probes are not that far apart and can definitely be explained by a difference in calibration for the most part. The small amount of difference we see in the pattern of the trend line could be explained by the variance in the temperature graph. Dissolved oxygen is a more complex reading to explain. Most researchers agree that it is a crucial factor in determining water quality [5]. The problem comes in with the fact that there are a number of variables involved in calculating dissolved oxygen (including temperature and barometric pressure) and the way those are fed to a probe can have a lot of impact. The probe we used in our sensor cluster did not account for barometric pressure at all, so that is one possible source of error. Also, we see that there is a fair amount of variation in our temperature readings which can be magnified when factored into the dissolved oxygen reading. Another thing to consider is the high variability in dissolved oxygen that can be found in aquatic environments in very small pockets. It is not surprising that we see the largest differences in the DO readings based on all of these factors. Even at that we did end up with a strong correlation between our readings and the function sonde at 0.807081. The percent error was by far our highest, though, at 16.4226. It should be noted that the downstream YSI 6290 v2 had a failure in its DO probe shortly after it was placed in the artificial stream so it did not provide any usable data. For the dissolved oxygen comparison we were only able to use the upstream YSI.
This poster discusses research on bringing real world applications for wireless sensor networks into the classroom and covers the use of a wireless sensor network (WSN) using the ZigBee protocol to remotely monitor an artificial aquatic ecosystem.
Relationship to this item: (Is Referenced By)
Tools / Downloads
Get a copy of this page or view the extracted text.
Bunn, Zac; Guerrero, Jose; Wolf, Lori; Fu, Shengli; Hoeinghaus, David; Driver, Luke et al.Bringing real world applications for wireless sensor networks into the classroom: Telemetric monitoring of water quality in an artificial stream [2012],
report,
2012;
(https://digital.library.unt.edu/ark:/67531/metadc155617/m1/14/:
accessed April 23, 2024),
University of North Texas Libraries, UNT Digital Library, https://digital.library.unt.edu;
crediting UNT College of Engineering.