 Hi, please introduce yourself. Okay, I'm Oliver from Berlin, Germany, from Berliner Sensor Technik. And we have developed and produced a full Internet of Things. So where does it go there? What is it used for? It is used to monitor the wastewater at Deutsche Bahn high-speed trains. So, waste from the toilets and stuff like that? Yes. And what is the monitor with it? So, it's pumped from the tanks and this pumping can sometimes fail. So there can be a problem with the pumps. And the flow of the wastewater is monitored by this device. And if there's something wrong, then it will automatically be diagnosed and sent wirelessly by LTEM with 4G. What is the monitor in there? What does it check for? It checks for flow. So whether it is a proper disposal or not. And the challenge in this case is that it's not simply a yes or no question. It's a question for a pattern recognition system because sometimes the flow goes slower and sometimes it's stuck and sometimes it's not stuck. And that is a difficult decision. And where is your sensor? The sensor is inside in this case. So it has a probe which you can see here. It's a little hole in a protected corner so it will not be hurt by the water. And it's reliable? It's mass production? It's already used? Not mass production but production. So there are 40 of them in use for over a year, 24-7. So they are permanently in use. And they have a battery life of more than 3 months. So with one battery charge it works for quite a long time. And it has a permanent connection to the wire. Is it possible to make smaller ones for vans and buses and stuff or is it best to use it in train? We could consider that. But there is a special challenge with the trains because the trains have a lot of trains and there are a lot of pumping stations. So currently there is not much transparency. So it's difficult to figure out whether they are working or not. With smaller units it's easier. And we hear the Zephyr booth. Is it running Zephyr? Yes. So what is the chip in there? There is Nordic NOF 916 which is a beautiful arm chip which works with integrated LTE-EM and NB-IoT. So it's integrated wireless in one chip, in a single chip. Other units have dual chips but it works in a single unit. And it's all operated by a real-time operating system. What's the advantage of using Zephyr? We have several sensors working in parallel. And they have to work at the same time more or less. So we have an NFC reader which is here. So it can detect the number of the sensor and the number of the tank and the train. Because if there is something wrong with the train then it has to know which train and which tank. And that's detected by the NFC sensor. And it has this wireless connection with the LTE. That's the second part of this multitasking system. And you have the sensor of course. You have an accelerometer. So several units at the same time. How long is the battery? Is it connected all the time? It's connected all the time. But still it works for more than three months. So we are planning to have a run time of 12 months. But right now it's three months. So it keeps charging or there's a battery? When the battery is empty it's replaced. And it's charged at another place. Because this has to be shockproof and waterproof. And it has to be permanently used. So we cannot dismount it in order to charge it. Are there other people doing stuff like this? Or is it very special? No, it's very new. So it's brand new. There's nothing like it. Because generally you have flow meters. And the flow meters detect the flow of liquid. But that's the kind of thing. That's the kind of signals that are sent by the senders. And normally flow meters they have to do with liquid. But this is wastewater. So it's very hard to tell whether the pumping process has completed or not. All right. Cool. And so it's a great show for you. And you're doing presentations. Yes, we are at the Embedded World Conference as well. And Jonas from Phytec and I, we have had a class today about Zephyr. And about pattern recognition. Which was quite popular. So a lot of people were there. It was great interest. And we really enjoyed it. The chip has AI? The chip has... We have a special pattern recognition. It's all built in software. So it's our own software. We do not have a standard toolkit taken from others. It's something that we built ourselves. So it's unique. You don't need to use one of these neural network accelerators? No. It would not be reliable enough for our case. Because it's not tolerable to have wrong judgments. It has to be a very strict rules according to the customer from Deutsche Bahn. So they have a very strict decision of what is detected and how. And this works according to their specifications. Cool. Thanks a lot. Thanks a lot.