 Well, hello, hi, everyone. Thanks for inviting me. What I'm going to do today is talk about my experience of sensor networks and some ideas for the future. So I started this 20 years ago, almost in 2002, and this is the front cover of Computer magazine in 2004, but they had a special issue on sensor networks. And we wrote the chapter on environmental sensor networks. I need to see that we were lucky to be highlighted on the front cover. This is our probe, which I'm going to talk about in a minute. And we talked about how there are three sciences that kind of you need to understand for sensor networks, firstly sensors, communications and computers, competing. And we talked about the challenges. So this is back in 2004 that you needed to allow sensor networks to happen. And in particular with sensor networks, the two sort of problems we highlighted were power management and radio communications, because most sensor network systems were kind of indoors where there's plenty of power and there's not much problem with communications. But when you go outside, that's when all the sort of problems occur. And then a couple of years later in 2006, we wrote a paper, you can see there, a review paper called a revolution in system science because what we suggested was that sensor networks are about to take off and they're going to be a really, really big thing. And we made some predictions about the future. And what are sensor networks, well, they are essentially you embed sensors in the environment, they send their data by usually by wireless onto the internet. So in theory, you have live or almost live data from the field, and that's the excitement about them that suddenly you can sense the environment from your desk. You can use it for hazard warnings, you can make predictions about events. You can use it for fundamental science. So that's particularly important in places like glaciology where it's very difficult to get to the environment in the winter. So you need something that's continually monitors and volcanoes are another example which hazardous. Together to have some concept of smart monitoring where you might sense more if the weather says there's going to be a potential and avalanche, then you could sense more. And so have some idea of kind of smart, smart monitoring. And we can see sensor networks as part of a continuum within the environmental sciences. So back in the 70s really we started developing logging where you have sensors. They have their data stored probably initially on magnetic tape, but then on SD cards and you go back and you collect the data perhaps once a year. But you've no idea whether it's actually working throughout that year. So you've got to wait usually a whole year with the fingers crossed. So the excitement of the sensor network is that you constantly getting the data back so that you have a system where you have sensor nodes in your environment. They send the data to a base station. We send forwards that data essentially onto the internet so that you can check and get your data back and that's the excitement. And then we move on to the next phase, the phase that I'm going to talk about that we're in now, which is the internet of things. And where we have both both ways. So we're not only having data coming back from the field, but we're also able to talk to the nodes in the field. And that we use internet protocols to talk to these so basically the nodes are like web pages. And they're connected directly. I'm just going to start with a couple of slides about pre 20th century. In the old days, I didn't have big data had small data, each individual, I didn't have many data points, but they took a lot of effort to get. And also because you were camping, you're really low tech, you really have very little chance of any power to your to your systems. And as you kind of move towards the 20th century, 21st century, so as you move towards the 21st century. You started to need power in the search. And this is particularly illustrated by by the, what's called gpr ground penetrating radar, which is a fantastic tool in glaciology, because it tells you how deep the glass here is where the rivers are relatively easy to use but very, very power hungry. But all of a sudden you need power in the field. And I want to contrast this with our early days and sense networks and you see it's a completely different environment and to work in. There's always last minute things that have to be done. You need power. You need somewhere that's dry for that for those instruments. Why did I get into it. I'm my basic research is in subglacial processes. Now the subglacial environment so that environment under the glacier. It's one of the least explored environments on earth, because it's so difficult to get down there and find out what's happening. And yet it's vital because it controls the rate of which glaciers respond to climate change, and it's vital in understanding the rate of which sea level rise is going to occur. And there's two ways you can do it. You can either wait for the glacier to melt away and look at sedimentology, or you can do what's called in-situ process studies. So you basically do experiments beneath the glacier. And during the 1990s there were a series of experiments, really exciting experiments that were done with wired instruments and the data was stored on memory cards. And so we went into this to try and sort of update this to try and put all the instruments into one probe, into a multi-sensor probe. And to send the data back via radio and have it as a part of a sensor network. So we installed these probes, they're about 16 centimetres long, either in the sediment beneath the glacier, the till, or in the ice itself. They send the data back to the surface, and the data is then relayed to a reference station which has main electricity, and then goes on to a server and on to the web. So this was the GlaxWeb system. We started this initially in Brickdale's, bringing Norway, and we then moved to Iceland. We then moved to the Cairngorms, which I'll talk about, a short trip to Iceland, I'm sorry, to Greenland, and then back to Iceland again. So this is Brickdale's bring where we initially started to work, this is it in 2001, 2002, and you can see that a minute we started to work there, it rapidly retreated until by 2006 we couldn't work there anymore. And although this is one glassy, this is pretty, although an extreme example, this is pretty typical of all glasses in the world. Almost all glasses are retreating, not all of them as fast as this, but certainly glasses are retreating. So after that we moved to work at Skella Ffalsjökla in Iceland, a little bit more of a stable glacier. But it was a great place to work similarly. So the systems changed over the years as technology improved, but this is the basic system again we used in Iceland. We've got the probes in the glacier, sending their data onto the base station onto some sort of gateway. In this case we were able at the end to have a 16 kilometer wifi link to a farm and then going into the cloud. And so we were able to get data back every day for two years was the maximum we got but over a kind of a 10 year period on and off. So this is the details about the probes, these are the frequencies we're using. We managed to reduce the frequencies as the years went by, and we're essentially measuring temperature, pressure, strain, conductivity and tilt. This is the probe diagram for those interested. So on surface of the glacier we have a base station and this in itself is quite difficult because of course the glacier is always moving, and in the winter there's a lot of snow. So everything has to be really watertight. On the base station we have, we have the links to the rest of the world, we have a GPS weather station and the power. In this case we've got a solar power and also wind power. This is inside of the base station box again everything else we absolutely watertight because any water in or completely wreck it. We used a number of different systems but this is one in particular where we had a Linux based base station. And then you need somewhere which does have main electricity and in this case there was a cafe kilometer away and which had again a weather station GPS Wi-Fi. At the final stage our site is about there, and we were able to send by Wi-Fi our data down to this friendly farm, where it was then put onto the internet. But it did have to cope with severe weather. As far as I should say that one of the kind of rationale from the beginning was that if we can get the sensor network to work in a glacial environment, then hopefully it will work anywhere. Just a thing to show you about the kind of communications adaptation. There are a few pictures of working in the glacial environment. I mean it's pretty hard to put the, to get the sensor network to work and you need a lot of people. He has asked with the GPR in order to get the probes into the glacier at all you need to have to drill a hole using hot water and use the hot water drill and this weighs 200 kilograms. In Norway we were able to fly onto the glacier with a helicopter but they don't have many helicopters in Iceland so instead it was driven on with a vehicle. So quite a lot of work to actually get the system to work at all. I've got some slides here. This is us drilling the hole and it takes about two hours to drill the hole down to the bed of the glacier. We put our probes in about a maximum of 70 meters and that was about as deep as we could get the signal back. And this is a video showing what the holes look like. So we have made this with our own custom made webcam going down, borecam. And you can see at the first the light is lit actually by daylight but as you get darker and darker you get deeper and deeper it gets darker and you can't. And eventually you reach a water level because you've melted the glacier and the water will stay in the hole. First time we went through it went completely black so we had to have a bit of a rethink and waterproof the camera a bit better. And this is it when it is very, I've edited this, this is close to the bottom of the borehole and you can see the hazy nature that's the sediment from the base coming into the water. And in this particular case we made a mistake and drilled where there was a river and you'll see that the image is moving. And that's because of accidentally hit a river. And you go through and you can see the base and it's actually very rare to see the base of the glacier because it's so difficult to get down there. We've got to move it a little bit for the focus but there's some till at the base there. And this is an image of us putting the probe in and there it goes down the hole. And what we do is that basically we've blasted away the till, we put the probe in and the probe that the till will then close back around the probe so that it becomes embedded. And this is a little graphic that we've got that we made from the tilt measurements showing the movement of the probe. When it first goes in, it moves around a lot because this is open hole because you've blasted away the till, but eventually the till will close around the probe and hold it in its natural position. And here it is now and from now on it moves, it moved a predictable amount but it is kind of set in, set into its position. So I don't want to go into the details about what we found but basically we're able to produce this first environmental sensor network in a glacier environment with our own custom hardware and software. But for the first time we were actually able to show how probes move, how plasts move within the till and this is a big, big question in glacial sedimentology. And also able to get these seasonal measurements of stick-sick motion and make some estimates about subglacial sedimentary processes, which again, very difficult to do and we're very proud that we were able to do this. So now I want to talk a little bit about Internet of Things. So here's this concept of Internet of Things, you can see that quote there, a world where physical objects are seamlessly integrated into the information network. So, and here's some examples on this slide of some sort of off the shelf things you can buy from Internet of Things. As you can see, they're mostly indoor type type activities, as I said before, it's much easier indoors, but we wanted to apply this concept to the outside world. We decided to try and do it in the Cairngorms in Scotland. So this is where our field site is on the Cairngorm plateau, and we were using this Internet of Things concept to look at some periglacial features on the surface and look at some Pete studies. This is a kind of image of a typical sort of periglacial landscape up there. We should actually choose this site because, while the problems we're working on, it's pretty tough to work there. And also, it's very difficult to get there outside of the summer. And sometimes things break down and it's nice to be able to visit them throughout the winter, which is very hard to do. So we thought that in Scotland it would be much easier to access some of the mountains and we chose a site where there's a road and a state road that goes up to the surface private road and we were allowed to use it by the land owner. Before you could get to the road, there was a ford. And unfortunately, every time it rains, it meant that the river rose, and of course in Scotland, it typically rains quite a lot. So unfortunately, we were actually prevented from visiting our site as often as we'd like to because of this access problem. But once you could cross the ford, you could drive to the surface. And it's right up on the plateau. So this is a view of the system, this is a planned view. So essentially we have a series of nodes again in the landscape. But this time each node has this global IP address. So basically it has its own website. These talk to each other to a local base station, which we call on this diagram, a router node. And these router nodes were approximately a kilometre apart. They talked to each other. They were able to talk by six lo pan down to the estate building and that was three a three kilometre. Reach and in the estate building was electricity. And so we were able to forward the data onto the internet via a satellite link. And this was set up so that we could add extra nodes into the system. And here's a picture showing our us in action. So these are these are these router nodes that I was telling you about. And here's some us inserting some instruments into these periglacial features. And here's people how they get up onto the plateau. So instead of making everything ourselves, we now have moved to these standard protocols. So each of these different elements are different parts of the system. But each of these now we're using standard a standard way. And together they are linking. We also did some experiments with web interfaces. We found that HTTP is unsuitable. So we use this co-act thing, which is basically the Internet of Things equivalent. And we use this operating system cantiki. So this is what one of our nodes looks like. You can see all the relative elements of it. And this is an overview of the sensor nodes. And we had two sets of sensors. We had some very typical of the shelf things. And then we developed some smart sensors where we were trying to look at how water pressure and temperature and tilt changed in the soil profile. So here's us deploying one. So this is one of our router nodes and this is the power source. This is an example where we're looking at changes in water level and one of these little lockens. We were able to get quite a nice data set. This is an example of how the different temperature throughout the profile going down from the surface. So the great thing about this system is that it did work. So what we learnt, well, we learnt that you can see there sub-giverhurt, low power radios are sufficient to provide the range. And that this 6.0 pan provides internet protocols everywhere. And we were able to use this co-app to provide standard interfaces. So essentially we managed to get internet of things system working up on the Cengorn plateau. And now let's move on to another system. And this is our web connected RTK DGPS system in Iceland. Now before I talk about that, I'm going to talk a little bit about how we got involved in this. Now interestingly, we were contacted by Formula E and they asked us whether we thought it would be, whether you could put a one of these cars onto an iceberg and track it. The idea being that because it was Formula E that it was supposedly green, we said we didn't think it was a very good idea. But they asked us if we could make an iceberg tracker. So ridiculously short amount of time, we made this tracker. And they put it on this iceberg in this location here in Greenland. And using Iridium satellite, we were able to track the iceberg on its journey as it headed off towards Newfoundland. And it traveled about 500 kilometers in two months before we lost it, presumably fell off the iceberg. But it was a good, it was a good sort of introduction to this whole tracker idea, which we've been toying with anyway. So in many aspects of environmental science, we use GPS is traditional GPS is of course are very expensive. So you can never afford to have very many. And most of them still store their data on memory cards because they are. It takes a lot of energy to have very long messages. So it takes a lot of energy to get the data back. So what we wanted to do was to make a much smaller one that was cheap so it didn't matter you could let it fall into a crevasse or fall off the front of the glacier. And which would send short messages so that we could get, we could get that data back live. So luckily there's a kind of new generation of GPS boards which allow you L1 L2 RTK GPS so that's very high precision GPS. There's a number of different companies but the one we went for was Swift navigation. So we built the system, which has this GPS system running on MicroPython, and we used a medium short messaging to get the data back so that's essentially a sat phone to get the data back. And so this enables us to, we were able to have short messages that could come back every single day and we were taking three measurements. Sorry, four measurements a day, which gives us a really good detail about glacier movement. And because obviously because it's a cheap system you can have more of them. So this is a picture of the inside of the unit. Again, everything's got to be incredibly waterproofed. And we've set it up in two adjacent glaciers in Iceland, by the Merkirachor, Fjallirachor. And the idea is that we look, adjacent glaciers should be behaving the same because the next one and other, but we're looking at how they behave differently. We're both retreating as many glaciers and in Iceland there's this additional thing that's since 1990. The lakes in front of the Icelandic glaciers have been growing much very quickly and that has caused the glaciers to increase their melt rate. And one way of understanding sublacial processes is to measure the glacier velocity. And you can either do that from satellite note sensing through UAV, repeat UAV surveys, which we've been doing, and also by using the GPS. And the thing about the GPS is that you have a low spatial cover, but you have very high temporal resolution. Whilst it's kind of opposite with the remote sensing, although the best coverage maybe is probably at the moment is about once every 12 days. You can have big coverage but low temporal resolution. So if you can have all three of these, then you really got a good understanding of glacier response to individual weather events. So this is the rover sitting on the glacier. This is a kind of adapted Icelandic version, which the Icelandic metalfish use, and we've kind of adapted that. You can see it's got its solar panel on and the antenna, et cetera. And this is the base station that sits on a marine in front of the glacier. And the data that we've got back is very impressive, which they show lots of exciting things about these two gases, which are completely adjacent but are behaving very differently. And the great thing is that we've got data from, we installed this in 2017, and it worked. We had to go back and change the batteries and update a few things, but it worked until last summer. We were unable to go back because of the COVID and were unable to go back this summer, so unfortunately they're still sitting out there, but the batteries have unfortunately run out. But we were able to get three years worth of data for this, and we're very pleased. We have been, we're involved in a initial experiment from Swarm, which is a nano satellite company. And they've given us the kit to try and change from iridium messaging to this, these nanosats. And that's something we're going to try, hopefully when we go back next summer. What do we found from, from this work? Well, we found that we developed this first sense network for glaciers using custom hardware and software. We learned about this IoT style sense network, which uses standards based protocols. And then we've developed this web connected RTK DGPS system. And although we've developed it for a glacier, you could use this in many other environmental systems and it worked for three years and we're very pleased about that. Right, well, we're back to the kind of the beginning, when will the revolution happen? So when we wrote that paper back almost 20 years ago, we imagined that there would be a huge take up of sense networks. And although there are, there is has been some take up, it hasn't been as much as we initially thought. We've recently just published this paper about sensor networks and geohazards in an ebook called Treaties and Geomorphology, where we looked at particular sensor networks and geohazards and did a review of those and there are quite a lot of sensor networks. They're kind of traditional ones. Interesting use of UAVs for hazard detection. And also this growing field of citizen science where sensors are perhaps being carried by people and sort of a different way of perhaps looking at sensor networks. So there has been activity, but maybe not as much activity as we'd initially thought. So challenges, if we come back to those challenges that we talked about at the beginning. So, in relation, batteries are still a problem. Obviously there's all sorts of problems about lithium batteries in the environment as well. We said, I didn't mention this, but we said at the beginning that you really want your system that doesn't destroy the environment, you want a system that fits into the environment that essentially mimics it. You want to make sure you are actually monitoring the environment, but you don't want to damage it in any way. The environment has improved. It's still a long way to go though. Radio communication has more choice. I've got a slide to talk about that in a minute. Scalability is still problems. Remote management is a lot more improved with an IoT system. Usability is still poor. We'll come back to that. Standardisation, again, improvements, but needs uptake. We see more awareness strung pod to protocols. And there's always problems of having kit out on the environment and the security in both sort of both means of the word. Something that we hadn't thought about, of course, was big data. I mentioned at the beginning that prior to working on sensor networks, I basically worked on small data, but times have changed and big data is everywhere. It's so important that environmental scientists have a training in how to deal with big data. Now, although we have fair protocols, there's still a long way to go about where this data is being kept and you can have access to it, all those sort of different issues, which is still a problem. So this is a little diagram showing communication systems that are available at the moment. Obviously, they keep changing over time, but this graph shows sort of power against range. And when you're designing an environmental sense network, you kind of need to know what range and power you have available and each one will be subtly different and have its own sort of positives and negatives. And I think there's, there's kind of two ways to standardised sense networks at the moment. You either have a system like this where sort of only this part of it is in the internet. So you can have one radio type protocol, or you go for this IOT stack where the whole thing has web and internet standards so you can have a mix of software and hardware. We think this is this is the way to go and this is the system we use in Scotland, but there are places where you can, you can only use this system. So conclusion, oh I finished it early haven't I, but conclusion. For this you need, you absolutely need a dedicated and multidisciplinary team. You need the environmental scientists to understand exactly what you're measuring because you've really got to try and mimic the environment. There's no point measuring something that isn't useful. You really got to try and measure it exactly and you've got to understand that environment. And then you also need the engineers and computer scientists because at the moment it's still at a kind of laboratory level, you really need that high expertise. Funding, you need funding, it's incredibly expensive to do any of this. Although we've talked about no cost, everything's relative. And you definitely need patience and good humour, it's very hard working in these environments, even if you're working in Scotland, which is not, obviously, as cold or remote as a glacial environment, you still need kind of good humour to get on because things go wrong, things always go wrong. It's all your way. So the future. As I said, I think we've come a long way in 20 years. There are sensor networks, there are a lot more than there were, but we haven't perhaps had the take up that we expected. But I think I still have faith that I think the revolution will take place. I think that we're at stage now where it's going to take off because environmental sensing is the future. The fact that you can sit in your office and monitor the environment is just such a benefit to all environmental scientists, to society in general, it's such a good thing. But I think it is going to take off rapidly now. And I think that's because there is more standardisation, we can make more web-like sensor networks and people are used to using the web and so this is not such a big step up for them. There are far more sensor networks than there were. I think that our big problem, what's holding us back at the moment really is usability. I think we've kind of made good progress with the standards, but it's just the usability that's perhaps holding people back at the moment. Well, I couldn't possibly have done this on my own and it's a really big team has been involved in the projects over the years and here's a photo of all the members in the team. So people, as I said, you need lots of different types of expertise, both environmental scientists and computer scientists, but there's so much that you can achieve. Right, thank you. Fantastic. Thank you very much, Jane. That was a really great talk. Delighted to see that you're using the Swift Pixie board. I backed that on Kickstarter when it first came out, so delighted that it's out in the wild. There's some great questions coming in. So I'll direct these to you in turn. So there's two questions, actually, and you started to allude to this in some of the slides towards the end. What are your thoughts on the deployment of non-recoverable sensors into the natural environment that then become litter at end of life and do any of your probes get collected or do they have to be left in the environment? Is there anything that we have to grapple with a lot in environmental science? So if you could give us some comments on that. Yeah, it would be lovely if we could have biodegradable sensors and that would be kind of the aim for the future. We don't. We can't recover our probes. We'd love to, but we couldn't. Because as you saw at the beginning, our glacier breakdown was being completely treated on us. At one point our kit fell into the lake because it melted so quickly and we were able to go out with BBC News actually to try and collect it because we thought we could rescue it. And we did find a little bit. We managed to get the base station and then we went out with a metal detector to see if we could find it around the edge of the lake, but we couldn't. Yeah, it's sad. I think this is something that we need to move into in the future to try and build environmental sensing systems that are less damaging to the environment, but we're not there yet. What was the second part of the question? I think I think you've covered that. Yeah, that's great. So yeah, something quite difficult that we have to deal with soon. Next question, are sensors for one environment, for example glaciers, really that different from another? For example atmospheric sensors? Well, again, this is, no, I think that it's, sorry, I forgot what I was saying. I was going to say before that at the beginning in 2002, the big thing was to have this concept of smart dust, the idea that you could have sensors that were so small that they could sense the environment. But we've kind of been against that from the beginning because we thought that that would be so dangerous in the environment that it would be like releasing microplastics into the environment. So we've been, I don't think that the future is smart dust at all. Our sensors interchangeable, that would be a nice thing to have. But I think that for most settings, because of the power management problems, you really need to build it specifically for that environment. You need to build a system which we had, if I go back to this slide, with the communication system, you maybe could have built a system which had different communication elements and would click in, you could perhaps, you know, change it to be able to operate on different levels and similarly you could with an Internet of Things system, you could make it change and what sensors it sensed. I think it would very much depend on what you were actually sensing because in most situations you've got to make it as low power and efficient as possible and it might not be efficient to have multi-use sensors. So it really depends in what environment you're using. If you have plenty of power and plenty of connectivity, then you could do that. Thank you. A question from Matt Lewis. High-lasdewed places experience little sunlight during the winter. Has this been a problem for your research, for example, data bias in monitoring or powering of sensors? What's your power management strategy for the winter? Well I suppose a basic power management is to make everything as low power as possible. I also had a very clever system where the probes actually stored their data on RAM so that if they couldn't talk that day, they would store it until they could talk. That worked quite well because there were sometimes where the weather was so bad that the communication system didn't work down to the bed of the glacier and they worked pretty well. In fact, they stored data almost four months so that worked well. Yes, there are obviously problems with power on the surface and that's why we had wind power which was pretty good because it was pretty windy in Iceland. The solar power is not that much used in the winter but it's okay in the summer. We actually managed quite well because our key strategy was as low power as possible. Obviously, some days it was just impossible. In general, okay. Next question. How is data from sensors embedded for long time scales calibrated to ensure accuracy? Do you have to do any sense of calibration? Yes, we calibrated all the sensors before they were put in the lab. I was going to ask, do you have an idea of the kind of drift on your sensors for long deployments? Yes, we do have an idea because there were other ways that we could tell, say like for instance, how thick the glacier was. Yes, we had a few parameters, independent parameters that you could get some idea of the drift. I was going to say something, I forgot. We also had a system where we put some probes into the till beneath the glacier because we're interested in subplacial processes but we also put some into the ice so we didn't drill beneath the ice. We knew for certain they were in the ice because we hadn't reached the till and so then we could tell because we knew some were definitely in the ice and some that we thought were in the till and we could see difference in the sense of reading. So you could really sort of see that there was that sort of binary difference and sometimes we actually found the probe, one probe actually popped out of the glacier even though it was 60 meters deep. It popped out because obviously it came out of the till into the ice till interface and zoomed up, shot up a crevasse and came onto the surface. And so you could actually in the readings, see a sort of intermediate when that when a probe is actually at the ice till interface which is quite interesting something we didn't sort of expect. That must have been a surprise. It was. I'm going to start with a question from Matt Fry. What do you think what do you see as key components to increase uptake of sensor network technologies are there opportunities to provide standard packages of kit and data platforms to make it easy for research to do this without a whole team, or does every problem need a different solution. And I guess this is particularly applicable to environmental scientists those of us who haven't got much background in engineering how do we increase the uptake of these technologies. Yeah, I mean that that is that is the key question. I mean at the moment, you do need a team, but the plan is, you know, that is what we want in the future we want everybody to be able to use it just like use a mobile phone. Just like you can take a GPS off the shelf we want it to be able to everyone to be able to use it. And as I tried to say in the talk, you know we're moving towards it these it standards are helping people move towards it. So we, we just need to somehow kind of jump over that that threshold into usability. And perhaps this is where this particular, you know digital environment can help with that that that jump that vital step that we need. Great. Next question from Thomas Turf and Gels. Hello Thomas. Do you think the launch of space based cellular broadband networks that could provide global 4G 5G connectivity will make the setup of sensor networks more affordable and increase uptake. Yes. I mean I think this is all. This is all part of it. Yes, I think that it will be helpful. I mean, this is sort of just anecdotal when we were in Norway you could stand on the glacier and your mobile phone works. We live in the new forest and our mobile phone doesn't work here so you know connectivity is absolutely vital for internet of things. So yeah it's going to be really helpful. That's why we're so excited as well to try out these nano sats and see, see how that's how that's going to work. Yeah that's going to be very very very interesting we're trying out a couple of couple of similar versions so it'll be really cool to compare notes on where we are in in a couple of years time. Question from Mike prior Jones asking, could you elaborate a little more on your experience of using full IOT stack model rather than traditional telemetry plus gateway model, bringing IP down to each node means a lot more computing power at the node and potentially higher power consumption result. Can you explain additional benefits of having IP all the way to the notes. So I'm just, I guess go back down to the, it's all about power really everything has been designed to be as low power as possible. The great thing about having them all having their own IP addresses is that you can communicate with them individually and you can add new ones into the system. We just just get we just managed to get them as as low power as possible so you use and if I just go back to this. Yeah so you make sure in these these layers each one is absolutely as as low power as possible because you're always struggling, always struggling with power so you've got an operating system that's that's low power, all these, everything is basically low power to keep it going. And it works, and we were quite, you know, amazingly impressed that it got across that that three kilometers is quite impressive. Yeah, I think that's really that's really cool and you know in a way, simplicity to have it IP all the way but only if you can make the power compromises. Yeah. A question from Dinjanta Busan Das, purely from an experimental point of view, have you experimented if there was any difference in data collected from the same environment using a different sensor network architecture. Oh, that's a big ask. Well, in the sense that I guess in a way, actually, yes, in the sense that that we were there quite a while, and that every year we kind of updated. Because as we developed because certainly with the glass web system we started in 2002. We finished basically in 2012. 2013. So that is essentially 10 years and during that time, you know, things became minute more miniaturized. We were able to have lower power. And so in a sense, yes, we were sort of different systems because just because we're in the same place. But it was really, you know, it's really hard and every time you have to update the technology that's a whole change to almost everything the hardware the software. So a lot of iterations of all those elements, which is quite, you know, time consuming. A question from me thinking about kind of data data transmission and packaging and trying to make your data as small as possible you talked a lot about this in the GPS. How did you manage to do it with the tilt sensors that I really love the video that you're able to make of the probe moving in the till. Did you do much processing on board in order to get the data small enough, or were you able to transmit all the data from the accelerometers? Yeah, we transmitted all the data and we did the analysis back home. Those early systems we wouldn't have had enough power or computing power to do the analysis there. We just, you know, we sent it all back. I think initially once every three hours and by the end it came back. It was recorded every every hour and then it comes back once a day. So that's why we only send the data back once a day because that's that's the really sort of high energy thing to do. And how were you able to send sufficient data in order to get that positional data? You know that that must have been a huge data transmission daily to get that kind of detail. How did you design the data packaging? I think they're all pretty much the same. I don't think that was any more difficult than anything else. It's all just about making the packages as small as possible. You don't send anything that's not vital. You just send the essential information. Thank you. Just an opportunity for any final questions if you could pop them in the Q&A box. And while I give you the opportunity to type those, Jen, can you tell us what's next? What are you going to? What's your next research goal with these these sensor networks? That was going to be my question, T. In terms of field work, I guess, yeah, other experiments planned. Well, I guess our next plan is to try and I guess we've got two different sort of sections. I mean, one, we've got two fields. I mean, first is to try. I'm just going to go back to my picture. But going back to the Internet of Things thing that we'd like to do out of system where you could change. You could have a sort of like multi multi sensing system, which you could use anywhere, which you could change the power of the communication systems that you used, and that you could change it from. From your desk, so it would be an Internet of Things system which had different communication systems depending on what you were sensing in the environment. So sort of multi sensor thing. We'd love to make that map that sort of on the one hand. And then our kind of second element is to develop our RTK thing, but to have it loaded by drones. And we think there's a big chance of using drones to deposit sensors to collect sensors to use the sensors themselves. And as drone technology improves, we can relate this to sensor networks. So that's that's what we'd like. So those are our kind of two avenues that we'd like to go forwards within sensor networks. Great. We'll be, we'll be watching this space and find a question from Thomas Robinson. How do you know when you have enough data and how do you decide how many sensor nodes to have. That's like how long is a piece of string, isn't it? I think that's that that's probably a philosophical question in the old days. That's why I showed that picture was that we had, we had, we had small data, but each individual point was really valuable and took our two hours to collect. You wrote it in your notebook, you then lose it, you kept thinking you'd lost your notebook. I think it moved to this big data where it was. It's not there's too much data. It's just, you do have to know what what's important. And I think part of the sense networks is because you have to design everything so low power, you don't collect anything at the moment that's not vital. That might also change in the future. I think you have to plan from the very beginning what is the bare minimum that you need to sensor system. When you, you can't sense everything, and you do have to put a sensible amount. And that is quite difficult. If you think about Victorian surveyors, they didn't have many points in the environment, but they were able to produce their maps. And as you go through time, the maps get better, but they're not that different. They're still the highest mountain. You kind of fill in the details. And I think with the sense network, again, you might not be able to pick up everything, but having that basic framework is a huge step forward. And there are many things in environments of science. So we know nothing about, I mean, I can negation environment. We know virtually nothing about the winter. And since we've been able to put in sensor networks and logging systems, suddenly you know what's happening in the winter. And the winter is actually a really exciting time in negation environment. But before we had these systems, we didn't know. So I would say when you build sensor network, you build for the lowest at the moment, you build for the lowest frequency, I mean, sampling frequency, because you want to get the data back. And there's always this element of fear that you're not going to get anything back. I mean, the first time, the first year we did our study, we didn't. None of the post work, nothing worked. And then we went back the next year and everything worked because we learned from our mistakes. So I would say you start with a bare minimum, and then you build from that to get a bit better picture, but we can always have a better picture the world, but we want to know a base level of what's there at the moment. Fantastic. That's a really good point to end on. Thank you, Jane. I really, really enjoyed that talk. I hope I pronounced it as well. A reminder that this and all the other webinars will shortly be posted on our YouTube channel. So do go ahead and check it out there or check out our web page digital environment org, where you can find details of the past webinars and the next webinar. We're taking a summer break now, but we'll be back on the 10th of September. Speaker is still to be confirmed for that. But in October, we'll have Scott Ensign and Shannon Hicks from Enviro DIY telling us about how to create your own sensor system and a load of tools that they've developed in order to enable people to make their own sensor networks and get that data onto the web. And then more polar stuff at the end of October. Shridhar Jarawak from the Svalbard Information Observing System is going to be telling us about satellite data applications in Svalbard and sensors that they use there. So please keep an eye out for our speaker announcement for 10th of September. Have a lovely summer break and we'll catch up in the autumn. Bye bye.