What does it take to build a global, citizen-led environmental monitoring network? If you ask Radu Motișan, the answer involves more than just good code – it requires electronics design, server architecture, and a strong commitment to data transparency.
Radu founded uRADMonitor in 2012 with a network-connected DIY Geiger counter in Romania, and it has grown into one of the world’s most significant crowdfunded sensor arrays. Today, uRADMonitor tracks everything from radiation to air quality for thousands of users across every continent.
We spoke with Radu about the journey from a single sensor to global infrastructure, and why he believes the best way to understand our environment is to measure it ourselves. We are excited to hear Radu’s perspective on what it takes to run low-cost, outdoor-hardened sensor hardware at global scale, and how a project born out of post-Fukushima radiation concerns became a blueprint for citizen-led environmental monitoring more broadly.
Welcome, Radu! You built the first uRADMonitor unit back in 2012, when almost no one was using the term “IoT” yet – a microcontroller that could speak TCP/IP directly to the cloud. What made you realize that background radiation and air quality were problems that needed a global, interconnected network rather than a single device, and what did it take to go from that first gamma-radiation sensor in Timișoara to a network with a presence on every continent?
Thank you, it’s a pleasure to be here.
My path crossed radiation monitoring long before it became uRADMonitor. For years, my blog, pocketmagic.net, documented DIY Geiger counter builds: high-voltage circuits, open-source firmware crammed into AVR microcontrollers with 2KB of RAM. So the interest was already there. What Fukushima did, in 2011, was turn a hobby into a necessity.
Watching that disaster unfold, one thing became painfully clear. It wasn’t just that radiation detectors vanished from the market overnight and old Civil Defense units reappeared at absurd prices (Those old yellow bricks). The real problem was the absence of a system. People were entirely dependent on official numbers, slow, opaque, delivered through channels you simply had to trust. There was no way for an ordinary person to see, independently and in real time, the radiation level where they actually stood.
So in 2012 I built a device that reported its readings directly over the internet, with no PC in between, the way it was traditionally done. A microcontroller that could speak the language of network protocols, TCP/IP, DNS, DHCP, and talk to a server in real time over an Ethernet cable. Today we call that IoT; in 2012, very few people were thinking in those terms. The device measured background gamma radiation and centralized the data on a server. That was the first uRADMonitor.
The realization that it had to be a network, not a device, didn’t come from me alone. It came from the technical community. I had no partners; I started alone, but I found extraordinary support on technical forums. People suggested improvements, and I implemented them as fast as I could. And it was those people who showed me the real idea: that many such devices, together, could continuously map the radiation background, something no single reading can do.
One number, in one place, tells you nothing, because you have nothing to compare it against. Value comes from density and continuity. Once you accept that, you stop building a gadget and start building infrastructure.
I made ten units, then a hundred, and it went on from there. That technical community became my first users, the early adopters. I owe them a lot – the people from elforum.ro and 4hv.org. Those domains don’t work anymore. It was a long time ago, precious memories from when I had time for hobbies.
The uRADMonitor product line measures many kinds of radiation, gaseous and particulate pollutants, and even noise. Why measure so many things when most environmental monitors only measure a few? What is the core philosophy driving this breadth of coverage that tracks such diverse inputs – radiation, pollutants, and noise—rather than focusing on a single environmental metric?
Because pollution doesn’t respect the categories we invent for it. A person standing on a street corner is breathing particulates from traffic, the NO2 from the same engines, possibly noise above healthy levels, and, depending on the geology under their feet, a certain radiation background. These aren’t separate problems experienced by separate people. They’re the combined environmental load on one human being, in one place.
Measuring only PM2.5 gives you a fragment of that reality. PM2.5 is also the easiest to measure well, which is part of why so many solutions start and stop there. And of course, there would be a lot more to add, but there are practical sensor limitations on what can be measured.
There’s also a technical reason, and it’s about correlation. When you measure many parameters at the same point and time, the data becomes far more than the sum of its parts. A rise in particulates alongside specific gases tells you it’s combustion; the same particulate rise with a certain particle size and no gas tells you it’s dust. You cannot do source attribution with a single number. Breadth isn’t feature-creep. It’s what turns a reading into an explanation.
And frankly, there’s an economic logic too. A network is expensive to build and maintain no matter how many sensors sit in each box. Once you’ve solved the hard parts, power, connectivity, surviving outdoors for years, a station on a pole, adding parameters is marginal cost for a large gain in value. It would be wasteful to put all that infrastructure in the field just to report one number.
But this breadth came at a real cost, and it’s the part people don’t see. Most competing solutions are essentially one thing: WiFi plus a PM2.5 sensor. We went the hard way. Over the years that meant supporting radiation, particulates, and a range of toxic gases, but also building across multiple connectivity technologies, WiFi, Ethernet, GSM, LoRaWAN, NB-IoT, even SDI-12 for specialized applications, because a sensor on a remote mountain and a sensor in a city building have completely different constraints. It meant multiple hardware platforms, from various microcontrollers, large code repos because everything has to be optimised for low memory and flawless automated operation, different enclosures, displays, power designs. Years of R&D, iteration, and failure went into that. It would have been far easier to ship one WiFi box that reads one value. But a single metric, on one radio, isn’t an environmental monitoring network. It’s a gadget. The whole point was to build infrastructure that works anywhere and sees the full picture.
To give one concrete example: cheap particulate sensors read high in humid air, water droplets look like particles. We addressed it by adding a small heater on the air intake, then validated it in tropical field tests in Vietnam in one of our products. That’s the kind of unglamorous engineering nobody sees, but it’s the difference between a number you can trust and one you can’t. And it adds up in time. This all does.
Radiation, especially nuclear radiation, has a bad name in the Western world even if we are all surrounded by it at all times. Why measure radiation directly when radon gas and ionizing smoke detectors are the only common radiation sources?
Because radiation is the one environmental parameter where public perception and physical reality are furthest apart, and that gap is itself the reason to measure it. People are either irrationally afraid of it or completely unaware of it, and both come from the same root: they have no data. We are surrounded by it at all times, yet almost no one can see it. A number on a screen, continuous and independent, is the most effective cure for both the panic and the complacency. When you can watch that the background is stable and low, the fear dissolves. And if something does change, you’re the first to know, not the last.
The idea that radon and smoke detectors are the only common sources is exactly the misconception worth correcting. Radiation is always there and it varies: with altitude, with the geology beneath you, with weather, with what building materials surround you. Granite, cosmic rays that climb with altitude, medical and industrial sources. It’s a genuine environmental variable, not a binary danger light.
And this is where the network matters most, and where uRADMonitor actually started. A single detector describes one point. A network establishes what “normal” looks like across a whole region, over years, and only against that baseline can you recognize an anomaly that means something. I started with radiation for exactly this reason: it’s the parameter people most need objective, continuous, independent data on, precisely because they can’t see, smell, or feel it, and because the official sources aren’t always fast or transparent. This was never about distrust. It’s about not having to depend on trust in the first place. You have the number yourself. This is the beauty of open data, of transparency, of information that answers to no gatekeeper. And I feel it helps us function better as a society.
This stopped being hypothetical for us. Together with Greenpeace, we deployed autonomous radiation sensors in southern Ukraine, near the Zaporizhzhia plant, precisely so that if something changed, people would have independent, real-time numbers instead of waiting on official statements.
In our work on Local Haze, we see a tiered system of environmental monitors: a small number of governmental “reference” monitors, a larger number of local government monitors, and citizen scientists with low-cost monitors. How do you see the role of citizen scientists and low-cost monitors?
That tiered system is real, and the mistake is treating the tiers as rivals when they’re complementary. Reference monitors are the gold standard for accuracy, but they’re few, expensive, and each one stands in for an entire city. They tell you the truth about very few places. Low-cost sensors are individually less accurate, but there are many of them, and density is its own kind of truth. When air quality changes street by street; only a dense network captures that.
The two make each other better, and this is not a slogan, it’s how the calibration actually works. You correct the dense low-cost network against the few reference points; the network then extends the reference station’s authority into all the gaps between stations. A calibrated dense network, anchored to a handful of reference instruments, gives you accuracy and resolution at once, something neither tier can deliver alone.
We’ve had this validated directly, in independent challenges and peer-reviewed studies using our open data. I am proud of very good results on accuracy – lab tested, an important achievement considering we are talking about “low cost” monitors.
And this isn’t theoretical: independent researchers pull our open data directly, sometimes with no special programming background, and publish peer-reviewed work with it.
But the citizen contributes something beyond a data point, something easy to undervalue: ownership. When someone hosts a sensor, they stop being a passive recipient of official numbers and start engaging with their own environment. Data you gathered yourself, about the air outside your own window, is data you’ll act on. That shifts environmental awareness from institutional abstraction to lived, local experience, and it preserves public trust through openness rather than authority.
The role of low-cost monitors isn’t to be a cheaper reference station. It’s to make environmental data dense, local, and personal, three things the reference tier, by its nature, cannot be.
Indoor smoke detectors are incorporating end-of-life timers, not just battery warnings, and HVAC thermostats are sending filter-replacement emails. Many low-cost sensors do not have long time spans, and the monitors are not always maintained regularly.
Is the maintenance of monitors something that the low-cost environmental sensor community should be working on by providing notifications and replacement parts?
Yes, and it’s probably the most under-addressed problem in the field. Everyone celebrates the deployment; almost no one plans for the years after. A sensor on a pole degrades quietly: electrochemical gas cells last about two years, particulate sensors drift, and the reading stays plausible long after it stopped being accurate. A neglected network doesn’t fail loudly, it fails silently, which is worse, because it keeps producing numbers people trust. And from a point, those are not accurate anymore.
The comparison to smoke detectors with end-of-life timers is exactly right, and it’s the model we followed. The device, or in our case the server, has to tell you. We recently deployed automated health notifications across the network: the server checks every device, and when a sensor goes offline or its gas cells pass their service life, the owner gets an email, grouped and rate-limited so it’s useful and not spam. Most offline cases are a two-minute fix, a power cycle or a router check. We also added real-time threshold alerts, so a sensor can actively warn when a pollutant crosses a dangerous level, not just log it quietly.
But notification is only half of it. You also have to solve the parts and the logistics: the replacement cell has to be cheap, available, and something the owner can swap themselves, without shipping the whole unit back across a border. We’ve seen the problem firsthand, a large share of any aging network is quietly running on expired components, still on the map, still trusted, and wrong.
I take this personally, because it’s a principle I care about: we stand by our products. That isn’t taken seriously enough in this industry, and especially not in our part of the world. Some of our earliest units are still running years after deployment. Imagine sensors going for 10 years now, and some of them are installed in harsh places like Alaska. Maintenance isn’t a product afterthought, it’s the line between a real long-term network and a field of slowly decaying boxes. Data you can’t trust over time is worse than no data, because people act on it.
Thank you for joining us today, Radu, and sharing your perspective!
For more information about uRADMonitor, visit https://www.uradmonitor.com/
About Radu Motișan
Radu Motișan is the founder of uRADMonitor and a technologist specializing in environmental monitoring infrastructure. He holds a Master’s degree in Computer Science from the West University of Timișoara. Radu has led the development of the uRADMonitor platform from the ground up, driving every aspect of the project—including electronics design, embedded firmware, server-side architecture, and the large-scale data infrastructure required to support thousands of global sensor nodes.
Photo credits:
Photos courtesy of uRADMonitor