I'm Lars, I studied high frequency engineering and am now working on my PhD in signal processing.

I always dreamed of having my own lab, but since I had access to well-equipped labs, other acquisitions took priority. Since the space of one entry is not sufficient to describe all my projects and activities, I would like to present another project here.

Again and again I came across Deep and Machine Learning. When I started looking into it, I quickly got bored with the examples because they either have images or statistics as a basis. As an electrical engineer, I thought why not use the output of a meter, spectrum analyzer, time signal, or S-parameter.

So I started collecting data with my NanoVNA, which has the advantage of giving me both reflectance and transmittance information.

With a first set of data, I want to classify the soil for our flowers, because this happens too often, so I just put two coaxial cables into the soil and actually measured changes (in the same pot).

Another experiment I use as a kind of radar to detect if a CD is included or not. I'm also playing around with RF fingerprinting.

I'd like to post the dats etchings on Github, along with documentation and examples. Also, I'd like to bring these kinds of measurements to lectures I'm involved in as a grad student.

However, with the VNA, I can only generate frequency ramps as with FMCW radar, but I would like to test other waveforms including step response. Furthermore, the computer interface of the nanoVNAs is rather cumbersome and poorly documented. Using SCPi commands would simplify the data acquisition and save a lot of time. All this would be possible with Keysight devices.

If someone would like to work on the datasets, I would be very happy, you can find me on the web:d.


Learn how to vote for your favorites.