Close

Visualization and Quality of Data 2 – The Ugly

A project log for SENSEation

Modular sensing platform for research

mario-freiMario Frei 04/11/2020 at 21:330 Comments

The figure below shows an example of quite low data quality. Again sensor nodes 4-6 don’t show much because the data is event-driven and the time gap between the events is usually too large. And sensor node 17 needed to be adjusted, just as in the example from the last project log. Sensor node 18 started only in March because the oil flow meter needed to be installed by a third party. There are two kinds of gaps; the first kind is where there is no data available for all sensor nodes such as at the end of January or in April. This kind of gap is caused by issues with the gateway. The other type of gap concerns only individual sensor nodes, while others continue to send data. Usually, these gaps are intermittent, however, there were cases where on-site intervention was necessary.

The gaps that concerned only individual sensor nodes could have been avoided by deploying more router nodes. The gaps concerning all sensor nodes were caused by different issues around the gateway, such as:  - The power supply of the gateway was unplugged by occupants. This issue required an on-site visit.

 - The antenna cable of the modem was disconnected (reason unclear). This issue required an on-site visit.
 - Unrelated issues on the web server: Once the web admin accidentally shut down our web server and in another instance, an unrelated script caused on overload on the web server.
 - Issues with the modem on the gateway. We suspect that in the case of bad cellular reception, the modem would increase the sending power, and therefore draw more current, which in turn causes a brown-out, which triggers a reset on the microcontroller. However, we were not able to confirm the issue. Sometimes, this issue was resolved by itself. Other times, an on-site visit was necessary. In that case, relocating the gateway would solve the problem. However, the Feather-Fona-Gateway does not have offline storage, and even with offline data storage, all data would probably have been lost anyway, because the microcontroller was stuck in a reset loop. Hence, we abandoned the Feather32u4 Fona. Instead, we opted for a gateway consisting of an Arduino Mega, Fona breakout board, RTC, and SD-card adapter. The Mega-Gateway showed much better performance with only very little lost data. Also, we redesigned the gateway housing in such a way that the antenna won’t be unplugged from the modem when opening the enclosure.

Discussions