 Thank you, Arvid. Very delighted to be here. Let me try to pull up my presentation here. If I put it into full screen, is it going to work for everybody? How does that look? All right, I'm going to go. So yeah, once again, thank you. Tom Rondo, I'm a program manager at DARPA. I've been there for six years. So quite a long time for those who know DARPA. We usually cycle in and out pretty quickly. But I've been able to use this time to advance a lot of the technologies that we use behind software to find radio and what we're terming here. I like to talk about programming the electromagnetic spectrum. And so what do I mean by that? Pull out an old tweet from a friend of mine, Ben Hilberg, that there are four fundamentals of the force of the universe. And we get to program one of them at home. So that's what we've been talking about, is manipulating this force, the electromagnetic force, as our core capabilities by building software, programming. So writing Python code that in 30 seconds, you can start actually transmitting or receiving over the RF waves and manipulating those signals through the Fourier transforms, which I are near and dear to my heart. You can go into that later, how near and dear. But we've made it easier by building programming frameworks around it. Specifically, the GNU radio project was one that I worked on for many years. It helped me get my dissertation finished and my PhD. And then I took over as the maintainer of the code base for six years before coming to DARPA and using the platform at DARPA to expand our capabilities and building new technologies behind it. So let's dive into what this programming of the electromagnetic force really means. So I'm a bit of a history buff, especially history of science and radio. And there's a much more detailed version of this chart, but I'm just putting it up here now for the sake of time to try to hit the highlights of the history of radio mixed with a history of computing science. And so what I love about this is how they, this interesting trends that you find when you look at the, when you look at these two timelines here, starting in the mid 19th century, you have some of the fundamental breakthroughs in both aspects of these two worlds. You've got the kind of the invention of computing science with Ada Lovelace and Charles Babbage, but you also have the threshold. Yeah. Yeah, I think your slides are not advancing. I wasn't afraid of that. Are you just seeing the... The first slide. Yes, okay. So if I... Ah, now it's good. Yeah, yeah. All right. Well, it ruins the joke of the tweet of the universe. So here's the tweet that I was referring to at the beginning. Yeah, so here's the timeline. This is where I really wanted to hit. So you get started with Ada and Ada Lovelace and their work and Maxwell, the breakthroughs in Maxwell's equations that really helped us understand empirically how the universe worked when we talk about the electromagnetic force that rolls in quickly. And what I love about this chart here is the way that I've tried to design it to show this dilation around the invention of the transistor and the transistor radio and computers. So you have early on some amazing work done in the RF field and the radio field to create the fundamental breakthroughs for how we understand how we engage with the electromagnetic spectrum. Meanwhile, the computing field, I think, there weren't terribly many theoretical breakthroughs. I'm not talking about engineering. I'm talking about theoretical breakthroughs for how we control these systems. The Turing machine, Alan Turing's dissertation and the work that he therefore then did in World War II was of course a big one, the creation of the first electrical computer, the ENIAC. But what's fascinating to me about this is the transistor, right? The transistor was built as a device to amplify signals. It was to replace vacuum tubes as an amplifier, not as a compute logic device. But it was quickly realized that if you can use this as an amplifier, you can also use it as a switch. And I'll also point to, I was pointing here to Shannon's dissertation as actually as a master's dissertation, considering one of the most impactful in the world, telling us how to use Boolean algebra to use switches to compute numbers, to use Boolean algebra and how that logic flows to compute pretty much any numerical computational solution. You put those together, you get to create the transistor-based computer. You also get to create the transistor-based radio. But when you really look at the histories here, the breakthrough in the computing world accelerated our ability to attack those problems and create a computational, it's a watershed moment for the work that we do there. And then you can just see again the invention. And these are creations of devices, but they represent some of the breakthroughs in how we use the transistor to manipulate information. Meanwhile though, I argue that we actually slowed down in a lot of those kind of fundamental breakthroughs in RF. We kind of just did the same thing for a long time by just doing better engineering using the transistor, but not really changing how we approach this. And by the way, I also have a variation of this theme for a radar and there's actually a very similar argument that's been made there until some of the breakthroughs that we had in the software to find radio worlds. And I point out here in 2001, the GNU radio space, that probably shouldn't maybe be more generically and more regenerative of the larger software radio world in general, but by really combining the use of the transistor as a numerical device for computation and the transistor as an RF device for amplification, you mix those together and you solve it to find radio. And that's leading us towards a highly flexible, highly performance as we're moving farther into the future ability to control the electromagnetic spectrum and to, as I say, program the electromagnetic spectrum to whatever we need to do. So if you look at what 5G is doing today, the parameter space that's available is massive. And then 6G is going to be bigger and how do we actually manage all that through computational interfaces and through the software to find radio models. So from there, I want to go into a couple of examples of programs that we've run out at DARPA in the past few years that have started to enable us to break this open even further. So I've already gone through a little bit of the history here. So I don't want to double tap that, except to say that by the time we created GNU radio, the Joint Tactical Radio System, JTRS was a military push. Vaanu came out of MIT with his dissertation and created the company, all going after this more programmable surface of the software find radio. So as we walk through the next couple of decades, we find that there's new approaches to the computational models here. So not just an X86 processor, the general purpose processor. You also had JTRS mostly going after the FPGA as to get that low size, weight and power requirements that we needed for the military. But now you have this space of heterogeneous computation. We've got lots of examples from there. One of them was a program we started a couple of years ago called the domain specific system on chip to really launch our ability to utilize heterogeneous processing because different processors are good for different things, for different math problems and software find radio represents a number of math problems. And so you've got this, what we came into last year, what we call the fourth generation of software find radio. And that was to take the programmability that GNU radio enabled us. So it was a very efficient programming environment, but not particularly efficient execution environment by just running on X86 processors or on processors. SDR 4.0 and the fourth generation is combining those to allow us to offload from that programming environment of GNU radio to the execution environment of heterogeneous processing. And so when we did that, we created the ability to abstract the difficulties of the underlying architecture, the plumbing of moving between different processing domains left that to the GNU radio scheduler and just allowed the programmer to decide which processor was required for a particular event. This is now baked into GNU radio. So GNU radio version 3.10 has a lot of the mechanisms here enabling this type of processing. And so you can see here before we did this down here, the processing was only about 40ish percents on average of what we were able to take advantage of in the computer. It meant that 60% of the time the processors were bored waiting for data. After SDR 4.0 by taking advantage of the data offload and management of the data flow, we get to spend a lot more time just focused on the processing. And that includes GPUs, FPGAs and other types of heterogeneous processing. So now we have a platform to both program and execute. So that's really exciting. And the fact that we pushed this out into the GNU radio open source community for us to continue that speed of development and exploration of new radio techniques. So I wanna end here and my talk by talking about another program that we're currently running at DARPA called the Open Programmable Accelerated 5G or OPA 5G. This one was all about how do we understand the utility of 5G for the military soldiers and Marines and warfighters? If we were to use this in the field, that could represent a significant threat from an attack surface. So we wanted to understand what that attack surface looked like. And to do that, we created a large core of government labs from Army, Air Force, Navy representation throughout this program, looking at what are their equities and what are their concerns. And then on the other side, we had the development of an open source software implementation of a 5G radio access network. So both the UI or user equipment, the handset and the base station are now fully an open source implementation of these for 5G. This is gonna allow us to explore and measure and monitor every aspect, every layer of the protocol stack. It's gonna allow us to reconfigure these devices much more quickly to understand, again, what the challenges are and what the design of our defenses are gonna be by having visibility, having an open programmable system. So we released the first implementation of this using the non-standalone 5G in 2021. We will be releasing an SA version. So we'll have both UI and a GenoDB SA standalone coming out this month. So we'll continue to accelerate that work and produce capabilities that are going to be far more accessible beyond what we're doing just in a Starper program from that defensive side and enabling us to grow the community of knowledge, of users, innovation that can sit on top of this 5G open source stack and continue to accelerate our innovation. Now, I do wanna mention we're also running a program. I don't have a slide on this. I just thought of it today because we're doing a program review. Another aspect of the future of programming the electromagnetic spectrum is using beamforming, spatially diverse diversification and orthogonality so that you can reuse spectrum and you can put more capacity out there. So big deal in 5G, it's gonna be a bigger deal in 6G. What we're doing in this program called TRIAD, Tensors for Reprogrammable Intelligent Array Systems, or sorry, Tensors for Reprogrammable Intelligent Array demonstrations is to build systems that can demonstrate the use of GPUs and tensor programming to manipulate the beamforming of an array of antennas. There's gonna be some exciting work coming out of there. Much of it is actually being developed as open source software and will be pushed into the open source community as well. And so with that, I will end my talk and open up for questions. All right, thanks, Tom. So let's see. Oh yeah, I think if you stop sharing, we can import the questions. There you go. Excellent, so we do have a question, it's a broader question, which says, how do you handle security? I mean, obviously that's like a 10,000 foot question, but I think specifically it was probably referenced in the latest development on the radio side. Yeah, so security is a huge thing. When people ask, are you gonna secure the system you got out? What do you mean by secure? It's a multifaceted dimension. So let me try to quickly hit a number of things we're doing. From a hardening of the code and a cybersecurity perspective, we are working with best practices, we're building the standard governance and maintenance of the source code. And of course, working with Linux foundation from the OSSF, the open source security foundation, using as much of that as possible to create good, best practices from the community, continuous integration, continuous deployment, and of course, mixed in without a means continuous testing. So we're doing a lot of testing on the digital side, quality assurance and unit testing. We're actually doing a lot of testing on the RF side. So actually having analog interfaces to our SDRs so that we can actually explore a variety of aspects there as well. So that's one thing. The security on the RF side is we're looking at, again, standard taxonomies of attacks that happen on the RF space. We actually are using a standard Cisco model for this, bucketing to denials of authentication, denials of service, jamming type of approaches and building up that taxonomy and then trying to tackle the ones that we're hearing from our end users, the services, what equities they're really concerned about and then developing experiments and a testbed to specifically tackle those. And then once we understand that, we can figure out, is it implementation of the code or the device? Is it a protocol problem that we can work with the 3GPP to understand or is it a network configuration problem that we can work with either the service providers or if we're using it ourselves as a private network, we can work around. So those are variations on the theme of security that we're interested in. Wow, and each of these could go on for days. So that's a very good broad answer. There's one more question. Are there specific applications or use cases you're looking at, whether it's in the battlefield or in facilities in terms of scenarios that you consider for the overall ecosystem or architecture? Yeah, absolutely. There's a number of scenarios. And we're working very closely with the Office of the Undersecretary of Defense for Research and Engineering. So I'll just refer to it as the Pentagon project. So the Pentagon has launched a number of 5G efforts. If you've read about them in the news, you'll hear maybe some of the 5G bases or the tranches because we executed them in a couple of different tranches of funding. We're tied in closely with that community to look at what they're developing for their use cases on the bases. So smart depots, rapid redeployment scenarios. So how do you quickly understand the logistics train of I need to outfit a marine company, I need to outfit a ship and how do you actually move that quickly throughout? So those are some of the deployments that we care about here from just managing our resources and our inventory as best as possible. In the field, what I'm concerned about is it's too easy to reach for our phones. We all know that we're addicted to our phones. We all know that they're incredible devices to get its information and to connect us. We also say we wanna fight like we train or we wanna train like we fight so that you're used to how you're gonna behave in the field. You're gonna therefore fall into the trap of trying to use your cell phone and you want that. You want that information and access. If that's gonna happen, we need to understand what those challenges are gonna be and how can we utilize that to the best of our ability. So these are security issues that we have in the field. We're trying to make sure we understand them and can give our soldiers the best protection and understanding of that environment that they can have. So there's a direct relationship to our operations around the world and our ability to utilize these commercial devices for our ability to maneuver and survive when we do that. Wow. It's amazing how use cases move well beyond YouTube's and cell phones and web's and into IoT and machine to machine and all that. So it's built on fundamental architectures. Yeah. It's the exciting thing about 5G and future G work is moving beyond that service model of just getting data to a user. It's getting all that sensor data, all of that mechanism, all the mechanics that are gonna be enabled through a wireless connectivity and redeploying that for all sorts of scenarios. For us, we have military applications for commercial world, industry 4.0 type of approaches. Really good place to look at dual use of this technology. Very nice. And again, I wanna personally thank DARPA and all the research that has been going on, whether it's vaccines to 5G to security and you're really excited. The future is bright with smart people like you all. So I appreciate your talk and your insights. Thank you very much. Thank you.