FAU Center for Connected Autonomy and Artificial Intelligence
|April 14, 2023
This episode features a conversation with Dimitris Pados, Director of the Center for Connected Autonomy and AI at Florida Atlantic University (FAU) which, in partnership with Open Generation member Virginia Tech, PQ Secure Technologies, and Florida International University, is creating a universal radio adapter to enable a host of devices (IoT and other) to seamlessly connect and operate through both trusted and non-trusted 5G network infrastructure. Dimitris shares why this project is important, the driving force behind the research, and how the team plans to ensure point-to-point security of mission critical data as it traverses the 5G network.
00:10 | Ajit Kahaduwe
Hello again everyone. Welcome to another edition of Open Dialogue Wavelengths, which is our monthly video series focused on keeping you informed about the groundbreaking work Open Generation members are doing to advance 5G innovation for public good. This month we're honored to feature the Center for Connected Autonomy and Artificial Intelligence at Florida Atlantic University, also known as FAU. In partnership with PQ Secure Technologies, Florida International University, and Open Generation member Virginia Tech, is creating a universal radio adapter to enable a host of devices for IoT and other use cases to seamlessly connect and operate through both trusted and non-trusted 5G network infrastructure. Today we're joined by Dimitris Pados, who's the Schmidt Chair Professor for I-SENSE. Acting Executive Director and Fellow there as well. And he's the Director of the Center for Connected Autonomy and AI at Florida Atlantic University. At FAU Dimitris’ focus areas include communications theory and systems, signal processing and machine learning with applications to connect robotic machines. Welcome Dimitris!
01:22 | Dimitris Pados
Thank you very much for the invite. It's great to be here with you, then.
01:25 | Ajit Kahaduwe
Yeah, we're really happy to have you in our series of videos here. So the work you're doing it sounds really like a very complex research project. What was the driving force or impetus to create this initiative?
01:41 | Dimitris Pados
Yes, so the motivation behind this initiative, I would say came from the NSF convergence accelerator program that called for proposals to develop technology to securely operate, securely communicate through 5G infrastructure. Yeah, we had this project idea in mind for quite a while. But I would say that was the final push for us to to put this thing together.
02:08 | Ajit Kahaduwe
That's, that's really interesting how that came together. So when we think about all these different endpoints, why is the seamless connectivity between endpoints that are over both trusted and non-trusted 5G networks really important? What's what's the problem that we're trying to solve there?
02:26 | Dimitris Pados
Right. So 5G, 6G, next G, whatever promised to it all. They promised to connect everything – people things, computing platforms, robots, AI robots in particular when mobility is involved, connected robotics, that is vehicles from the ground and air everywhere. Seamless, secure connectivity is a must, to rightfully gain the trust of the public. So that's one of the main reasons that seamless connectivity between multiple users, multiple endpoints is important. Whether it's all the trusted or non-trusted 5G networks,
03:16 | Ajit Kahaduwe
I think that's a really good point because with 5G networks we have mobility. So the endpoints are moving to users are moving. They may be moving independently, the endpoints. And even if the endpoint is static, there's always this requirement for security, regardless of whether it's trusted or non-trusted. So being able to understand all those mobility use cases and static use cases, intertwined, I imagine is really a fundamental part of what what what you're trying to solve. So in that light, how do you envision trying to authenticate all these end users and devices, which maybe are in very different environments. They can be in a car. They can be in a house. They can be in the university and enterprise. They could be on a boat. How do you see handling all that?
04:01 | Dimitris Pados
That's that's exactly the idea. Sure, let's imagine a number of different users machines exactly, as you said, connected on the network. The natural thought there is to look for any indigenous native kind of fingerprint that you can catch. This can come from the RF from the user modem, for example. And definitely, in this context, machine learning or AI will, will have a big role to play to play in that. So to summarize, this thought, any fingerprint of any kind that we can grab, might be good enough. But naturally, the thought is to look for I think it's been coming from the front of each individual user more than how do you classify based on the RF fingerprint? That's another story. You can have the issue of machine learning AI plus classification approaches or anything else that might come to mind, but this is what we're looking at basically, RF front fingerprinting of individual user modem devices.
05:12 | Ajit Kahaduwe
Right, that's actually a really interesting idea. Because traditional security authentications, really based on unknown security parameters that are passed which can also be faked. But for everyone who might be listening, you may not believe it. But radio transmitters, even though they come out of factories they have a unique signature to them. And that's been used in different applications. So it's interesting that now you're looking at using that as part of the authentication scheme to add on to all the other security layers that are going to be used. So when we then look at ensuring the point to point security for, let's say, a mission critical data, as it goes across the 5G network, we talk about the authentication, but how are you going to ensure that part of the the user plane is protected?
06:01 | Dimitris Pados
Right, yeah, this question gets really, really to the heart of the project. I think that's it and what makes us so excited about it. We're going to acquire IQ samples of the user signal, at the minimum necessary sampling rate, then are going to post quantum encrypt the IQ samples themselves and push them across the 5G network, without ever attempting to decode. Actually, now you can think of this as an antenna remoting kind of operation. Or maybe what I could, personally, a spectrum transfer over five G, so we take a chunk of RF spectrum from point A, and we get it to point B, not over the air, but through the 5G network. So this is the heart basically of this project. I would say.
06:56 | Ajit Kahaduwe
That's even more interesting. In fact, I thought the authentication part was pretty interesting. But thinking about sending an IQ signal directly from point to end point without doing any processing would make it very difficult to decode. And you can still have all the other security layers and tunnels around your data as well. And that would be, yeah, I think in the post quantum world, some, we need to have unique new abilities to be able to protect against data being exposed through the power of quantum computing. So the scope of your project, to me, seems very large. How has your team been able to divide the scope into a work plan and across all the different partners that are that are working on this?
07:41 | Dimitris Pados
It's really a truly naturally forming team, with some of the best in the business in wideband RF front design, AI fingerprint classification, RF circuit development RF systems from development and IQ acquisition, then we have FPGA accelerated post quantum secure encryption and data hiding. Each team member here takes care of one or two items in this row, and the whole is created seamlessly so it's really a naturally forming team across those different types, I would say. And, of course, the objective here is to reach the end result within a certain time period, as soon as possible that is.
08:35 | Ajit Kahaduwe
yeah, it's I find it fascinating that the number of different disciplines you just mentioned in his team to be able to do something it's I think this is kind of the new world, we're moving into where it's not just a radio problem or an encryption problem. You mentioned AI, you mentioned all these other team members that and competences that need to be invoked to be able to create a solution now, which is kind of the world we're moving into. So you mentioned a little bit about timeframe. But how long? I'm pretty excited. How long would you think it would take to complete these research requirements? And when do you think we may actually see this technology going into a commercial use case or by even the military?
09:17 | Dimitris Pados
Yeah, we're looking at maybe a two-and-a-half-year horizon from today depending on funds, availability, of course. And interestingly, it is not only the Feds, that is NSF and DoD, that fund or will fund this this effort. There's also significant I would say direct interest from the industry. And in that way, this project is one of those few maybe that make me really confident that it will happen and we'll get to the end user soon. So I would say in two and a half, maybe three year’s time we will have a working platform along the idea of this project.
10:02 | Ajit Kahaduwe
All right, so that platform we can then use to try an experiment and, and that'll be the proof of concept that can then be used for further commercialization.
10:13 | Dimitris Pados
So right. Think of this proof of concept as either some sort of a prototype that take a point A that thing reads the spectrum does not decode what's happening. Again, it's not our business, to decode to get down to the bitstream, we sent the IQ stream, we sample it at whatever level is appropriate for us. We encrypt for security reasons. And then we push through the 5G network in kind of a hidden way, the different ways to do that you may embed the IQ stream. In other media, you can just you know, follow traffic of other users over the network and get them across. So a successful prototype, which is something that really takes spectrum IQ samples from one point and gets those IQ samples to another point. And then you do whatever you want to do with, those IQ samples. We have some ideas about testing this device in certain, you know, 5G private networks. And that should be done within within those two years. First two years of the project.
11:20 | Ajit Kahaduwe
That's incredible. Yeah. And I was thinking that this kind of technology requires the capabilities of 5G to make it possible. We couldn't have done that in, in earlier iterations of cellular from GSM to wideband, CDMA LTE.
11:37 | Dimitris Pados
Exactly. It's amazing, isn't it? Right, we now can get maybe spectrum from one place to another or over a network?
11:44 | Ajit Kahaduwe
Yeah, yeah, It just opens up the possibilities as we continue to move forward. And as we end at even as we're talking about 5G now, and what this could do 6G is already being worked on. And there, there's even more capabilities around the corner and at the latter part of the decade. So let me close with a last question. Florida Atlantic University joined Open Generation last fall and I want to understand how can your participation in Open Generation help you when you want to meet the goals of this research project, and others at your school and the Center for Connected Autonomy and AI in the future?
12:22 | Dimitris Pados
Ah well as Open Generation likes to say, “to reach our 5G future faster the key is collaboration,” right? So this project that we discussed, I think, is a good example of collaboration, researchers and developers must stay connected exchange ideas, discuss successes and failures. And this is what MITRE Engenuity Open Generation 5G Consortium is all about. So I think for us, it's been an excellent platform to exchange ideas and truly build collaborations.
12:56 | Ajit Kahaduwe
Yeah, it's Yeah, I think speaking for the consortium and all of our members, it's great to have universities because you're working on use cases that aren't in the commercial world yet, right. And being able to have the facilities like our private network to be able to test and all of our partners that can also give feedback, and work with the university students in your program is I think it's a it's a nice match for everyone involved. And we even get a preview into what's going to happen in the future assuming that the commercialization is viable after you go through all your research steps. I'd like to thank you, Dimitris, for joining us today and taking this time to talk about what you're doing at the university. And also for those who have been watching this series, this is our second in our series, and we will continue to bring interesting speakers to talk to us on Open Dialog Wavelengths that will give you some insights into industry and academia and the what's going on in the 5G world that will be coming in the future. So thank you.
13:59 | Dimitris Pados
It’s my pleasure. Thank you.