This software-defined sensor system downloads its user application securely from the Internet when it starts up.

Technically this first system predates Defsystem as a company, but it was done by the founder of Defsystem and started our journey into sensor systems.

Lab setup

We are a bit reluctant to show this first picture as this setup looks a bit more homemade than our later systems. It consists of a controlling computer with screen and buttons, a temperature sensor that can be submerged in liquids and a PH sensor.

This PH sensor is a simple analog sensor of low quality. It is not accurate, not fit for 24/7 measurements, needs frequent calibration and has a limited probe life. In all, this is DIY consumer hardware, and one should not expect more than +/- 0.5 PH accuracy. If at all. We do not use it anymore and do not recommend using it.

(In more current systems, for example to automate a hydroponic system, we used much more reliable and accurate industry grade hardware by Atlas Scientific, USA, to measure PH levels and also electrical conductivity (EC) of liquids. Of course, that comes at a different price tag but the combined accuracy is about 100 times as good as the consumer system, it is fit for 24/7 use, needs much less frequent calibration and the probes have much longer life expectancy. But more on this in a future post…)

Software-defined sensor device

As described on the page Our services and solutions, this system is an example of an software-defined sensor device:

The system is programmed in a high-level programming language which allows for fast development and short implementation cycles. Everything but the bare networking and connection part is stored in the back-end data system and fetched over the air by the sensor system when it starts up. Thus, the sensors are defined by the back-end system that delivers the source code and configuration and thus only then assigns the task. This allows for fast iterations without the need for re-programming of the devices on site.

In the first screenshot you see how it downloads the different program components and then it announces to start it.

Welcome screen

Once started, the system acquires the current time via network time protocol (NTP) over a Wi‑Fi network, and presents a very reduced user interface which can be controlled with just three buttons.

Calibration screen

With the command "cali" one can check the active calibration and also start a manual calibration process of the PH probe. The system guides the user through the process step by step (not shown here).

Selecting the solution under test

With the command "SuT" one can select the solution under test that is going to be monitored.

Measurements and data storage

The monitoring is started with command "start" and this will do repeated measurements of PH value, temperature and current time and store it to the database backend system Dydra by Datagraph GmbH as JSON data with REST calls over HTTPS using Wi‑Fi.

Delivered system

This is the more compact m5stack computer used in the end, instead of the bigger one that looks a bit like a handheld gaming console in the previous screenshots.

Dydra web site

The measurements are stored in the Dydra graph database by Datagraph GmbH, publicly available as dydra.com.

Visualization and analysis

Finally, the measurements stored in Dydra can be visualized and analyzed for example in Observable notebooks.

• Back to main page