A recent project got me to thinking more deeply about the term “Internet of Things”. We’ve come to know mostly what it means — the continuous flow of sensor data to the cloud facilitating insights — but therein is the rub. It’s the “data”, not the “Internet”, that we’re after.
In the case of my recent project, Internet was not an option. My sensor device is underwater in a remote location where Internet is cost-prohibitive. The cost-friendly alternative is to collect and cache the data for later collection. In essence, I achieve insights without the “Internet”. My solution is a “Data of Things” solution.
What did I build?
I built an Arduino-based device to measure water depth. It is accurate to 1/100th of an inch. It records depth on preset intervals so that I can review the changing depth over time. Time is maintained by a real time clock, accurate with within less than a second per year. It works by measuring pressure at the bottom of the water. Most importantly, data is logged on a microSD chip rather than streamed over the Internet. Thus, one might call this a “DoT” device rather than an “IoT” device.
Yeah, I know, “It’s not waterproof.” I overcame the waterproof thing by inserting it into a ziplock bag (three actually). The bags were then inserted into a five gallon bucket with holes in the lid and 40 lbs of concrete.
Why did I build it?
I built the device, five devices actually, to prove that a man made causeway is disrupting the flow of currents thereby eroding a beach, silting a harbor and causing nitrogen to be retained in a river. By measuring the height differential of water between the bodies of water on both sides of the causeway, the potential current can be estimated.
The causeway was initially built in the 1920’s to connect Gooseberry Island to the mainland. It was fortified in the 1940’s to facilitate access of military equipment to submarine lookout towers (still existing). It wasn’t long before residents began to notice changes to the currents and later the beaches. By the 1980’s, stones were being uncovered on a beach that had historically been known for pristine sand. Meanwhile the Westport River harbor now required active monitoring. Sand was accumulating in sandbars at the river’s entrance and buoys had to be moved as the accumulating sands grew and shifted. Then, in the 1990’s the harbor was dredged by the Army Corps of Engineers. What had once been a natural harbor was now under threat of closing up.
Another problem began to surface, algae blooms were becoming routine in the river. Algae consumes excessive amounts of oxygen in the water thereby starving the ecosystems of the wetlands. Testing showed that the algae was thriving on increased nitrogen levels in the river water. The source of the nitrogen was increased use of fertilizers, farm and residential, running off into the river water. But there is a second problem — the river is not flushing nitrogen at the expected rates. With each tide cycle, large amounts of nitrogen-rich water return to the river rather than flush away. The same redirected currents that have eroded the beach and silted the harbor also seem to preventing the river water from adequately flushing.
Why isn’t the Town or State doing anything?
Town Selectmen are working with the local Watershed Alliance to tackle these and other problems but resources are limited. Recent attempts to get state funding for a focused oceanographic study of the area were eclipsed by higher priority issues in other parts of the state.
I am working with the Town Selectmen and with the local Watershed Alliance to develop overwhelming evidence that will raise the priority for the next funding opportunity.
How did it work?
So far, it has worked amazingly well! The first prototype was dropped into the river for a single tide cycle on July 7. I had programmed the device to log the data in JSON, pictured below.
A plot of this data reveals the expected tide cycle:
Unexpected, but very interesting, is the temperature data that was also collected. It shows a 90-minute offset between tide and temperature. Said another way, the tide peaked 90 minutes before the temperature dropped to its lowest reading. The difference is mostly the result of the distance that the sensor is from the entrance of the river where the fresh, cooler water comes from.
The next step was to run the experiment for a week on the remaining battery charge. I collected 13 tide cycles before the battery ran down.
This test was interesting because it shows the diurnal tide, larger tides following the moon and smaller tides opposite the moon. It also shows the flood tides of the full moon around July 21st. Interestingly, the full moon was on the 19th. And it shows an interesting arrhythmia in the temperature, deserving of future study.
In a third test, I dropped four cloned sensor devises together to prove that the data is aligned. The variance averaged less than 0.001 inches computed from the hPa pressure readings.
The last step, deploying the devices at the causeway, is currently underway. I dropped them in on Saturday, 7/30/16, and plan to pull them on Saturday, 8/6/2016 and I will publish my findings in a follow up article.
Data of Things
While this experiment was unable to take advantage of the Internet for transmitting data to a cloud, it still achieved the same result. Therefore, it’s not the “Internet” of Things that matters as much as the “Data” of Things.
No doubt that the Internet makes life easier … no need to paddle out to retrieve a sunken device to get your data!
And there is no doubt that the IoT revolution has made devices more accessible and cost-effective than would otherwise be the case. Not only are devices such as Arduino, Raspberry Pi and Beaglebone available with numerous add-ons, but there are also numerous derivative devices designed to reduce size, cost, and development labor.
Data, not the Internet, is the ingredient of insights.