 Good day everyone. My name is Ngoni Wombashora. Today I'm going to be presenting my topic and the topic name is leveraging next-generation mobile networks for drone telemetry and payload communication. So just to start a little bit about myself like I said, my name is Ngoni and I grew up in Zimbabwe and moved to South Africa to pursue my tertiary education where I did my Bachelor of Science in Mechanical Engineering until 2020 and went on to do my Master's in Electrical and Computer Engineering, which I'm on track to be finishing soon. So yeah, today as the topic said we are leveraging next-generation communications and the key words in that are drone and communication. So where does this lead us? I'm going to start with a brief overview. I'm going to start with drone communications and what options do we have there? We've got a direct link. We've got satellite. We've got ad hoc network and we also have mobile cellular networks. So your direct link is essentially just your remote control communicating directly to the drone. Yeah, an advantage of this is it's relatively easy to set up with nothing complex there and it's a simple link of although it does come with some limitations and the biggest of those is line of sight. So if your drone goes a bit too far or maybe let's say behind a building or cheese or it's in a dense area with a lot of interference you are most likely to lose control of your drone. The other option is satellite communications and this gives your drone internet and communication access via the satellite and although that is good satellite is you've got remote coverage so that's an advantage over the direct link but the drawback of using satellite communications is that you have you've got very complex bulky and heavy equipment on board the drone. So yeah, you run into your swap constraints, which is your size, weight and power constraints. So because of that it tends to be very expensive, very expensive drones and so on. The other option is ad hoc networks and ad hoc networks are bespoke networks that have been set up for like a specific task. The drawback of this when you're using it with drones is that they are again they are for a very specific task and it's like an almost hecky way of setting up a communication to your drone and yeah, the drawback of this is that you run into like very complex routing protocols and ad hoc networks are not transferable to other devices and other use cases as well. So setup time tends up to be really high but yeah, they are quite robust networks. And the other option we have is mobile slash cellular networks. And this is essentially what we are going to be dealing with today. So mobile networks, you have got global coverage. They are available anyway, although you do like they aren't there in some very remote areas. But just like your cell phone, whenever you're moving from city to city, from country to country, wherever you are, you've got network coverage on your cell phone. So because of this, we are going to dive a little deeper into mobile networks for drone communications. So yeah, what are the problems? So understanding the problems with respect to drone communication. Like I had mentioned before, we've got our line of sight limitation. Yeah, if your drone goes too far, you lose control of it. We've got our uplink and downlink data rates. And this is highly dependent on the drone use case itself. When you're just command and controlling it, pretty much just flying it. You don't need very high data rates. About 200 kilobits per second is good enough. But let's say you are the streaming HD video that that becomes a big issue. And you want a communication system that's able to handle those high data rates. Latency is also an issue. Let's say you're flying a drone in real time, you're going to want it to react to your input as quickly as possible. You know, and it also depends on some use cases. Let's say you're delivering medical equipment and or you are doing drone racing. Latency does become an important factor. The other factor to consider is coverage. And yeah, pretty much depending on the use case again. Let's say you're doing farmland mapping or you are doing drone entertainment at a shopping mall, farmland mapping would require very high coverage. Whereas drone entertainment at a shopping mall, it's it's quite a localized thing. So yeah, because of these problems here, like these four issues, these are the four main issues. And yeah, what this ultimately leads to is you've got a limited number of use cases for drones. So yeah, just to mention a few use cases. I've put these use cases according to some of the limitations that we run into. And some of them, the first one here being coverage limitation. Like let's say you are spraying plants and you're on a farm. You're doing pretty much drone farming or power line inspection. Yeah, over here, this is a height constraint. So yeah, if you're doing those first two use cases there, your drone is flying pretty low and it's got an altitude of about up to 10, 15 meters, whereas compared to upper airspace inspection. Your drone is high up in the air and one might be testing, let's say carbon dioxide concentration in the air or yeah. So with regards to this, you've got a height limitation and you need some communication systems that can accommodate for that. When you're flying 10 meters in the air is compared to about 300 or 400 meters in the air and yeah, hotspot areas, you've got stadium shopping mall. That's your aerial entertainment, which is a hotspot area. You don't need much coverage there, whereas when you're doing farmland mapping or logistics and transportation, you need to be able to communicate with your drone from a long, a longer distance. And yeah, so with regards to data rates, like I mentioned before, command and control transmission is not demanding in terms of data rates. 200 kilobits per second is enough. Whereas when you want to do 8K HD video streaming or AR VR streaming or video capture, yeah, you need you need rather about one gigabits per second for AR and VR for 8K video streaming. You're looking at about 500 megabits per second. So that becomes a strong limiting factor on the on the kind of communications you need because it needs to be able to sustain those high data rates and be reliable at the same time. And yeah, in terms of there is also some use cases that are highly dependent on on positioning and accuracy. So for like aerial surveillance, let's say yes. Yeah, flying your drone in the one survey a specific area, you don't need much accuracy there. Let's say about one meter or even two meters is good enough for you to see what's happening in a certain area. Whereas when compared to automatic charging, your positioning accuracy needs to be, you know, maybe even within 10 centimeters. So these are just some drone use cases. And yeah, some of their limitations that come along with them as well. Yeah, so just some images there. You've got your aerial entertainment. You've got drone deliveries of medical equipment. In the picture here, you see Mark Zuckerberg. He is demonstrating how Facebook just added live streaming from every device. And yeah, he's waving at the drone there and it's directly streaming, of course, like that would require high data rates for you to be able to stream high quality videos. And on the right side, you've got your automatic charging pad. So when your drone is landing, they needs to be pretty accurate on where it's going to land and yeah, make making sure that it doesn't miss the landing pad there. So your GPS accuracy accuracy needs to be really good. So why did we go with mobile networks? The first reason is mobile networks have an almost ubiquitous accessibility. Like I mentioned before, they're everywhere, right? Wherever, wherever, wherever you are, they've got coverage. They've got superior performance as well in terms of data rates and latency, which you are going to get into a bit later. They're highly scalable and they're based on evolving standards. They're highly interoperable and they're backwards compatible. What I mean by this is that wherever you are, let's say your phone has got 5G and you know, you're connected to your mobile network service provider via 5G. If you reach an area where that connection is not strong enough or 5G is not available, it refers back to 4G and it refers back to 3G and so on. So, yeah, they're backwards compatible and they're everywhere. And for these reasons, we're going to look to dive a bit more into how we can leverage mobile networks for drone communication. And with regards to mobile networks, you are going to be specifically focusing on 5G. And the reason for this is that not much experimental research has taken place over research in terms of using 5G mobile networks for drones. And, yeah, 5G has come with a lot of buzzwords. And, yeah, for this reason, I've chosen to look at 5G for drone usage. The first one there being network slicing. And network slicing is essentially when your phone connects to a network, the network kind of picks up the kind of quality of service that you want from the mobile network and it creates a slice or like an instance of that network and then connects you there. So it's not like 4G, let's say, when you when you connect your quality of service is essentially what everyone else is getting or any other device that connects there is getting the same quality of service, which is which helps differentiate things. And we can leverage this in terms of drone usage because, for example, let's say I am doing a video call, right? I would probably need very high throughput as compared to latency. It's not as important. So I'll probably get a slice of the network with with preference to throughput as compared to latency. So, yeah, two things are ultra reliable, low latency, massive machine type communication. So 5G, the 5G core network itself is is capable of interacting with other machines using a specific protocol. And it also comes with enhanced mobile broadband. They pretty much make better use of the available frequencies and the frequency channel itself. And we're going to get into more about that. This one is my personal favorite and this is 3D beam for me. So the best way to describe this is using this picture there. As you can see, you've got your network tower, your 4G network tower there. And what this essentially was is you've got your one antenna pointing in some certain direction, some are unidirectional, but everyone is getting the same beam. So if you are further away, you're going to get a weaker signal and so on and so forth, whereas 5G instead of having one antenna, you have what they call an antenna array. And through the process of interference, what these antenna iris do can pick out your location where you are and concentrate a beam towards a specific user equipment. So that means you've got a more robust communication line in that regard. So that's really interesting because drone communication, for drone communication, because a drone is flying, it's a high mobility device and it's flying in the 3D space and you always need to have that strong communication channel also considering the interference that would come from your propellers and this 3D maneuvering as well. So the other one is 5G spectrum and frequency utilization. So 5G has made better usage of the spectrum and the frequency utilization and one of the things that it does that is via numerology. So what is numerology? Numerology is a technique that also is used to really bring out that ultra low latency. So in mobile networks, we have a thing called a radio resource and what a radio resource is is pretty much an OFDM signal, OFDM standing for orthogonal frequency division multiplexing. And yeah, it's a way of propagating data over the air. And what it does is that each radio resource has got a chunk of 12 subcarriers and let's say in the example here, we've got a 15 kHz subcarrier spacing and a one millisecond signal that takes one millisecond to transmit. Let's say it's some piece of data. And yeah, mobile networks up to 4G LTE, this is essentially what they use. And now with 5G, what they do is that you can change or increase the number, the subcarrier spacing that you use. So you can use 15 kHz, 30 kHz, 60 kHz, all the way up to 240 kHz. And as you can see in the diagram, what it means is that that one millisecond that was essentially used to transmit 14 OFDM symbols keeps reducing in turn. So 15 kHz subcarrier spacing as compared to a 60 kHz subcarrier spacing, you have reduced the latency from one millisecond to 0.25 milliseconds. So using this 5G is making better use of the frequency bandwidth it has and also really drastically reducing the latency. Hence that's where the ultra-reliable low latency is coming from. So yeah, so that's a brief introduction and background. And what am I going to be talking about today? So today I'm going to be talking about the implementation of an SDR-based open source 4G 5G network for drone communication. And SDR stands for Software Defined Radio. It's essentially a piece of hardware equipment that kind of implements all of the hardware that's required for radio technology for mobile networks. So instead of having your big antenna tower and with its amplifiers and attenuators and modulators, all that hardware is now implemented in software on the Software Defined Radio. And yeah, essentially we get a relatively cheaper piece of equipment that can essentially do the same thing as a radio tower. So the way I'm going to do this, I'm going to start with a 4G test bed. So these are indoor test beds that I've set up, a 5G test bed and then I'm going to talk about the drone setup itself. And finally I'm going to talk about how I've been able to implement a 5G mobile network for drone communication, essentially. And whilst I'm doing this, I'm going to be giving an insight into the kind of performance you would expect from a 4G mobile network, a 5G mobile network as well. And yeah, these insights are going to be comparable to your commercial mobile network providers. So yeah, what is a mobile network comprised of? So here we've got an image of our indoor mobile network set up and it's made up of a core network and an E0B. An E0B stands for Evolved Node B and EPC stands for Evolved Pocket Core, which is the core network of the mobile network. So the easiest way to think about it is that the core network is like the big end and your E0B is your front end. So in the core network you've got HSS, MME and a few containers there and your HSS pretty much handles all the subscriber data. So every SIM card that's supposed to connect in the network is saved and your MME is pretty much handles authentication. So whenever you connect to a mobile network it goes and checks on the HSS database. If this user with this specific number and this SIM card IMEI is allowed to be a part of the network. And you've got your base station which is the front end and your SDR there. We are using an Eters B210 SDR from new instruments. And yeah that essentially becomes your your mobile cell phone tower essentially. And the user equipment we're using I was using a Huawei P40 Lite for this and a Raspberry Pi Hat, a 3G, 4G and 5G capable Raspberry Pi Hat. Which is just a dev board which comes with a SIM card module and a SIM card slot. So you can just program your own SIM card according to your core network specifications. You program your SIM card, you program your core network and yeah you just insert your SIM card and it now your user equipment now connects to your mobile network. So from this we carried out some tests and what are some of the key performance indicators that we realized. We realized a throughput of 75 megabits per second on the downlink and 30 megabits per second on the uplink. A latency of 16 milliseconds and an RSSI of minus 51 dBm to minus 25 dBm. So just for context for RSSI which is your received signal strength index is just a comparison of the strength of the transmitter signal and the strength of the received signal. And just for context an RSSI value of minus 50 dBm in mobile networks is considered really good. And obviously zero dBm would be the ideal value, would be the best value. And yeah minus 50 dBm RSSI indicates a value that your mobile network is pretty much like commercially viable in that in that regard. And from then on we went to set up a 5G test bed. You'll see that our 5G test bed is quite similar to the 4G test bed because that's because the 5G test bed is a non-standalone test bed. A non-standalone test bed is essentially a mixture of 4G and 5G and I'm going to explain that how that is because in 5G we've got non-standalone and we've got standalone. So a standalone 5G test bed is essentially quite similar to a 4G test bed but instead of having a 4G coordinate work we've got a 5G coordinate work in the 5G base station. So here in non-standalone mode we use our 4G core network, we use our 4G base station and we use our 5G base station. So a 5G base station is added on there and what this essentially means is that we've got two cell phone towers and the way this works is let's say your phone would look for available mobile networks that it can connect to it will find this one to connect via 4G and once it has connected via 4G your 4G base station would receive user information user equipment in information and if your phone is 5G capable it would then transfer that connection on to the 5G base station and hence now your phone is directly connected to 5G and communication is done via that. So yeah what are some of the key performance indicators? We have 90 Mbps on the downlink so the downlink is from your base station to your user equipment and on the uplink we've got 30 Mbps a latency of 5 ms so when you compare that to the 4G latency that's less than half it went down from 16 ms to 5 ms so yeah that's also like that low latency that I was talking about when it comes to the implementation of the 5G mobile network and we've got the same RSSI reason being that channel communications are the same transmit power for the base stations are the same as well and just for comparison I've compared my test bed to some of the commercial mobile network suppliers here in South Africa and yeah so for our test bed you'll see that our 4G test bed is performing much better than a commercial mobile network in this case I was using MTN whereas we've got a downlink of 75 Mbps and it's compared to the 40 Mbps in the uplink the throughput was the same and you'll see that the throughput in the uplink also for the 5G was essentially the same and the reason for that is that the uplink the limitation does not become what network you are using or the channel conditions themselves but the limitation is now based on the hardware in your mobile network device so in this case in our phone the hardware is the limitation because the antennas in the cell phone can only transmit a certain amount of maximum power they can only use a certain type of modulation so there is on the downlink more mobile networks use OFDM and yeah orthogonal frequency division multiplexing or yeah multiple access OFDMA orthogonal frequency division multiple access whereas in the uplink they use OFDMA SC and that's orthogonal frequency division multiple access single carrier so much like that numerology I was talking about on the uplink mobile phones only have the capacity of using single carrier whereas on the downlink you've got more compute power more electrical energy and more transmit power so that's where you realize that even though we have changed a lot of things on the uplink it's too pretty much the same because the limitation is now the hardware used to implement that on the mobile device so yeah you can see the 5G obviously with better performance than the 4G there 5G test bed with a downlink of 90 megabits per second approximately 90 megabits per second as compared to the commercial one which is 252 megabits per second and yeah the reason for this again is we're using an open source we're using open source open interface to implement these and they're still currently in the development stages although they have managed to implement much of the stuff for the non-standalone desk they are still working on improving the throughput state although yeah we did have better latency and that's probably to do with the channel conditions themselves this is an indoor test bed and was done indoors whereas yeah with the antennas pretty close to the phone and all of that so that would affect the latency as well so yeah let's move on to joint communication we've covered mobile network so okay what do we do now that we've got our own non-public mobile network running right let's now go and fly the drone using that and yeah I'll just start with a bit of setup so you've got a typical drone there this is one of the drones that we used and yeah you've got your motors you've got your GPS we're using a year two GPS you've got your electronic speed controllers you've got your propellers you've got your battery and what we care about the most here for us in terms of communication is our telemetry module and the cube orange which is our flight controller it's an open source flight controller px4 pxhawk code cube cube pilot and yeah so essentially your direct link is the most common one you just fly your drone with your remote controller and yeah you can put payload data in there but the telemetry module would be used so a telemetry module is essentially just an antenna on the drone you would have an antenna attached to your laptop as well and the telemetry module there it's using eight eight six eight megahertz frequency and that's where you would transmit either your command your control data or like your command and control data as well as your payload data so like sensor information die die directly to your ground stations for so for the ground station we're using mission planner or cube ground control these are all open source drone ground control stations so yeah so another slide just a bigger picture there so yeah how do we make our drone a mobile network user equipment the way I went about this is by attaching an an onboard companion computer onto the drone and we use the raspberry pi 3b plus for this and yeah just attached via usb to the autopilot itself so now the raspberry pi can send commands to the autopilot and receive commands and receive commands or control information from the drone so the autopilot px4 which is your cube pilot your autopilot is essentially now you're now able to get access to it using the raspberry pi from then on we went to attach a pirate so the pirate is on top of the raspberry pi and yeah the pirate essentially is what makes the drone part of the mobile network itself the pirate comes with a sim card that can connect to the mobile network so yeah so that's your setup there this is the drone that we used um and just to zoom in there you'll see that we've got our flight controller there at the bottom and this case here on top is our raspberry pi attached to the pirate and yeah we've got our five antennas there um essentially for mobile network communication and yeah so yeah what does the full picture look like the full picture looks something like this we've got our 5j mobile network here which is everything on the left which you will see that we went through except that we only added mission planner so the core network itself and the mission planner computer are the same computer they are running on the on the same computer and so yeah so on the right side that's our mobile network um yeah and on the left side we've got our drone and with the pi with the pirate there so the pirate can go ahead and communicate to the mobile network and yeah all data is moved to and from so in terms of drones um there's a communication they use the communication protocol they used is called mavelink and that's almost a standard communication protocol used for open source drones and uh yeah that's what we were using here and we're using mavelink router and mavelink router is just a piece of software that enables you to route to listen to the autopilot so over here on the left on the left hand side here if you look with what a mavelink router container which is dockerized um it listens to the px4 and and and and sends that information to a specific ip address and yeah anything that comes to this ip address as well it forwards it to the autopilot so mavelink router they and within the core network we have added into what is called the application layer of the core network we've added a container a dockerized container they called mavelink traffic generation and what that does is essentially essentially also just an instance of mavelink router so when anything comes to the core network it's directly routed to the mission planner application or the q ground control application which is your ground control stations and yeah that's this is essentially it's this is how we we we get our drone um to communicate with our ground station via the mobile network so you'll see here we've got our red lines then that's essentially how traffic would move to and from the drone to your ground control station um the red solid lines being your 5g so from the piet antenna to the 5g base station to the core network and and to your mission planner application and the red dotted lines those are your that shows the channel communication for the 4g so the good thing about this is like i said mobile networks are being backward compatible in the fact that our mobile net net net networks has got our mobile network has got um 5g available as well as 4g so whenever it's flying whenever your drone is flying and you have got your 5g communication channel deteriorating or the channel conditions are not too great it will always back it will always revert back to 4g and 4g because of the frequencies that we're using um 4g is operating at a lower frequency than 5g and like the communication channel is is less uh susceptible to um to attenuation and uh yeah 4g pretty much has got a longer a longer range of communication as well so it will always interchange between the between the two depending on which one is the preferred or the better channel to use so yeah that's essentially it and what we've been able to do um what i've been able to do with regards to this setup i have been able to pretty much uh command and control the drone so i can arm and disarm the drone i can tell you to do an auto mission um yeah i i can i can do all sorts of things it says essentially there is a full and proven communication to the drone using this and uh yeah it is it is great and yeah currently what we are what i'm working on is uh going onwards we're going to do some bit of flight tests um so your the flight test will include both the 4g and the 5g non-standalone mobile network and we're going to be measuring some drone communication key performance indicators and what this essentially will be is that um um you know is using um a mobile network that much better and is it worth it uh throughout like uh you know is there a future way and as it stands with our indoor test beds that we have set up there definitely is a future and the drone industry can definitely um gain a lot from what mobile networks have to offer and then yeah ultimately this will pave way for the 5g stand alone test so the reason why we even use the stand alone 5g mobile network is because yeah it's it's still under development and although it does exist um yeah the connection will only last a few seconds and um stuff that i mentioned there like network slicing and beamforming and uh the low latency using different numerologies uh that's still currently being implemented and yeah it's this it's the it's the same it's even in the commercial scene as well um commercial mobile network suppliers are currently only deploying the non-standalone version but yeah so this um this set of experiments and vision will enable us to pave way for for 5g stand alone tests and really um leverage mobile networks for drone communication and yeah just to mention yeah it would be good to see how from my from my current tests which i've currently run i even put the data there because i'm still replicating those tests and um yeah making sure that i've got reliable data but yeah you can see that sometimes like when your drone is flying and they say it's making a sharp movement because of the current draw of the propellers you see that even your throughput is highly influenced by that by that interference like your throughput might be it an average of 70 megabits per second and it just drops down to zero like when you're making a sharp turn or something like that so it's all it's all in interesting data to see and yeah i can't wait to really um feel into these test flies and look into that so yeah that's me for today um yeah just my acknowledgement i'd like to thank alfred peace loan foundation uh for sponsoring my master's degree and send tech solutions for providing the radio equipment that i've been using for these experiments thank you so much thank you for your time and yeah thank you for your interest in my in my in my presentation as well if you need to know more um this is my name there and i'm available on linkedin and yeah which would be good if you have any questions i would have wanted to make the presentation in person but yeah there were some visa difficulties and um yeah uh the visas were not processed quickly enough but i would have been likes to be there in person and see everyone especially after seeing um the number of people we had booked to attend this lot today have definitely been good to see all of you um in person and uh yeah talk about this and see what you have to say but if you've got any questions or need further clarity um you can always um look for me via that link and um yeah definitely get back to you thank you so much