EXECUTIVE INTERVIEW: Multifunctionality for military sensor systems, defining AI at the edge
StorySeptember 09, 2024
Multifunctionality at the component level is enabling multispectral, multifunction-type sensors, Jake Braegelmann, Vice President of Business Development at New Wave Design, told me in a recent McHale Report podcast. We discussed how artificial intelligence (AI) algorithms are impacting designs and discussed the proper definition of AI at the edge. Braegelmann’s definition: The edge is where the sensor collects the data and AI can enable the filtering of that data in real time to get actionable intelligence to decision-makers. We also covered the U.S. Department of Defense’s (DoD’s) modular open system approach (MOSA) mandate, being aligned to the Sensor Open Systems Architecture, or SOSA, Technical Standard, and more. Edited excerpts follow.
McHALE: Please describe your responsibility at New Wave Design and your experience in the defense industry.
BRAEGELMANN: I’m the Vice President of Business Development at New Wave. That team incorporates the sales team, the marketing team, and our product owners, so the people with the responsibility of aligning the development efforts, our product development efforts, with what the industry is going to need, along with our Chief Technology Officer, our CTO, so we have that core team as part of bringing our products to market. My background is in computer engineering, specifically FPGA and ASIC [field-programmable gate array and application-specific integrated circuit] designs with high-speed serial, high-speed data transfer interfaces. Formerly, I did work for Lockheed Martin in their airborne mission computing team, and I bring that same experience, along with others at New Wave.
McHALE: I know New Wave plays a lot in the radar, electronic warfare, ISR [intelligence, surveillance, and reconnaissance], and other sensor applications within the defense industry. These applications depend on the signal process and solutions that companies like New Wave [put out]. What trends are you seeing from your military customers in terms of requirements for sensor applications?
BRAEGELMANN: I love the question. Everything New Wave does, or almost everything New Wave does, is really big-data-centric – wherever you have sensors digitizing the real world. Radar, signals intelligence [SIGINT], EO/IR [electro-optical/infrared] sensors, wherever you’re digitizing the real world and then processing that data.
What are we seeing in that space? Bigger and bigger data aggregators, data concentrators, lots of sampling – high-rate sampling. What that all leads to is more and more data being moved, being processed, being subscribed to. Different applications subscribing to all this different data from different sensors.
A trend within that is really the multifunction apertures. What I mean by that is where you see sensors no longer being specifically radar, or specifically SIGINT or radar-warning receivers.
[When] you look at the current generation of [systems] deployed, you have discrete functions or discrete sensors for all these different modes of operation, or solutions, on these platforms. What you’re really seeing a trend towards is multifunction, that a given aperture has to serve multiple needs through really intelligent design and very capable electronics so that it can act as multispectral, multifunction-type sensors.
McHALE: Would you say that FPGAs have gotten so sophisticated that they are enabling that multifunctionality, such as the ones from AMD Xilinx, and, of course, Altera [recently spun off by Intel].
BRAEGELMANN: I think the FPGA domain is certainly part of what’s enabling this. It is the FPGA, the programmable logic silicon itself. It’s also multichip packages and what you’re putting either into single-die or multi-die in package from RF to optics to processors to memory to the FPGA fabric itself, or programmable fabric itself. And security too.
Getting that all into a single package has allowed things like what they call staring in the RF [radio-frequency] domain, where you can look at very wideband at an open aperture.
The FPGA lines from the companies you just mentioned are really heterogeneous computing chips. [Having] all these different elements in one is enabling a lot of the multifunction capabilities we see demanded from the DoD.
McHALE: Today, you see AI everywhere you go: AI is in your delivery service, it’s at the store, and wherever it is, it’s always “at the edge” but I’m always asking: at the edge of what? Which edge seems to depend on the market or application. The edge in the military world from an application point could be defined as to where the enemy is engaged, whether it’s in the cyber domain, space, air, sea, land. But how do you look at it from a sensor perspective? I attended a conference in January where you spoke, and I thought you nailed the answer to that question. What you said, if I have it right, was that the edge is where the sensor collects the data, and AI can enable the filter[ing] of that data in real time to get actionable intelligence to military decision-makers. Do I have that right? And could you expand on that?
BRAEGELMANN: Yes. It is one of my favorite topics to talk about. I appreciate you bringing it up when you talk about the edge. It has become such an overloaded term to some extent. And as you said, you know [with] an edge definition, perspective matters a lot. You know, what perspective are you looking at the application or the concept of operations from to call it the edge. If I had to broadly characterize the space we work in, it’s high-performance embedded computing. If you look at that space, and all of our peers in that space, I think we would all probably say that the edge, from that perspective, is wherever the sensor is digitizing real-world [data]. So analog data, whatever is happening in the visible light spectrum, or maybe some RF spectrum, wherever that’s getting digitized into bits and bytes. That’s the edge – wherever the sensor is interfacing with the real world, and that’s the soonest place you can make decisions [from] that data. So, whether you’re looking at something that flies or drives or a surface ship: that’s all kind of the same model. At some point you’ve taken real-world data and digitized it. That’s the edge.
But then the question becomes, how close to that edge do you process? And often people will say, well, push to the edge as far as possible. As you said, AI/machine learning (ML), there are some advantages to going all the way to the center as close as you can, but I will say there are other tradeoffs to be made. At times, you may want to not back up all the way into the cloud. I [don’t] mean process in the cloud, but sometimes you want to be able to aggregate multiple sensors from a geometric perspective [as] multiple sensors may be coming from different positions or different pointing angles on a given platform or different domains. Maybe you have a EO sensor and an RF or a radio-type sensor. You might want to be able to aggregate that data, at least locally, on a platform and then run your AI/ML or other high-performance computing applications right there.
In general, I think people would still mostly call that the edge in that while you’re on the deployed platform, you’re on the thing in the field, or in the air, or whatever it is. But I don’t think the edge, as far as all the way up into a single sensor aperture, is necessarily always the solution. It can be at times, though, and certainly AI and ML are part of that. AI and ML algorithms now and into the future have the capability to use data in ways that we can’t imagine today. By giving these multiple data streams into the processors that are going to host these algorithms, you give the best odds that future [systems will] develop and succeed in those environments.
McHALE: New Wave is a member of the Sensor Open Systems Architecture Consortium, also known as SOSA. Your company develops products aligned to the SOSA Technical Standard. What does it mean to be SOSA aligned? And how does the SOSA approach as an example of a modular open systems approach, or MOSA, benefit the warfighter?
BRAEGELMANN: We are SOSA members and proud to be so. We do believe there are benefits to the warfighter, benefits to the entire ecosystem that we operate in, which ultimately benefits the warfighter as being SOSA and MOSA aligned.
At a very high level what does it mean to me at New Wave? Being SOSA or MOSA aligned generally means you’re going to design, develop, and manufacture to open standards, and even more so open interfaces. This doesn’t mean that every card has the same capability, the exact same chipsets, processors, and optics. I think sometimes that can be misconstrued. That is not really MOSA’s goal, necessarily. I think the real goal is open interfaces, whether we’re talking electrical things like power signaling or networking standards like Ethernet or PCI Express or data-interface standards, or whether we’re talking thermal and mechanical. It is so that you truly are designing to open interface standards such that you can change capability out, upgrade capability, develop with the expectation of future capability, and actually have those things insert into a system without a total redesign or a major redesign. It includes software interfaces, too, as there are applications where containers that are often used [and] segregation of different applications such that you can expect certain interfaces from a software perspective.
I think it benefits the warfighter in many ways, but there are two key ways: One is speed to the warfighter. How quickly can new capability get to the warfighter? Whether that’s application development, whether that’s new silicon at the chip level, whether that’s maybe new thermal techniques and power techniques. [Designing to open standards] can quickly get those things get to the warfighter.
The other one is also technical in nature, but it’s actually also heavily supply chain in nature. One of the big benefits that I think can get overlooked in MOSA [strategies] and the SOSA [approach] is without them, you’re relying on a fairly vertically integrated delivery of a system. The sensor aperture, the sensor processor, and the sensor-processing algorithm all kind of have to come from a vertically integrated supply chain, because they all have to know these proprietary interfaces, proprietary standards to bring those pieces together. Again, algorithms, processing hardware, sensor, aperture hardware. In an open system with open interface standards, you can not only envision a supply chain, but realize the supply chain where your application developers, your sensor-processor hardware, and your sensor hardware itself can at least theoretically be separated and interfaced and pulled together, which provides for a more dynamic supply-chain environment. Someone could develop the algorithms, develop the applications that are ultimately not maybe the manufacturers of the sensors themselves, or even the sensor processors. I do think that has benefit for the warfighter and they can get a more diverse set of capabilities and options to actually bring to the field.
McHALE: How does your prime-contractor experience at Lockheed Martin help when going to work with a COTS [commercial off-the-shelf] supplier like New Wave?
BRAEGELMANN: That’s a good question. I’m not the only one at New Wave that has a big defense prime history. A variety of us come from a similar background. I do think it helps. The easiest way to say it is that “we get it,” we get that these programs have very long tails you’re going to need support for long periods of time, not just technical support, but operational support, maintenance. We understand the life cycle. The other thing I’d say, is we know how hard it is. And by that I mean a prime contractor is buying a chassis, buying computer boards, buying interface cards, getting software from somewhere, and they’re putting it all together. That’s hard. Integration is still really hard and a prime contractor has to live that in spades. And we know how hard that is. What that drives at New Wave is real, shoulder-to-shoulder work. We know that we’re selling a standard product, but we still understand that standard product has to work in your system, in your environment, in your application, and that integration is not easy. What that drives inside the company is a closeness to the customer and a willingness to really get in there with you and really work with you on how you need it to work, where can it be improved, or [where] more performance can be eked out.
McHALE: Let’s look forward a bit. What disruptive technology innovation do you see being a game-changer in the radar and electronic warfare world? Predict the future.
BRAEGELMANN: I can predict a little ways. We talked about it briefly in a couple of the earlier questions. But it’s worth kind of putting them together here: You combine the multifunction aperture, the desire, the need for truly multifunction apertures, with the disruptive technology that is going to be chiplets or heterogeneous chips.
I’ll put a definition on that. When you start seeing optics – true, high-density, high-bandwidth optics – go in a package with your high-bandwidth RF ADCs [analog-to-digital converters], your DACs [digital-to-analog converters], with your processor cores, with memory and with your programmable logic. Imagine that all going into one chip, essentially. When you combine that world – again, big RF, ADCs, optics, processor cores, memory, and FPGA technology – all into one package and essentially have optical fibers out one side and maybe RF coax out the other side of this chip. Then you have multifunction aperture designs or deployments on platforms. Now you’re really disrupting what a current-generation platform would look like, whether [it is an aircraft, ground vehicle], surface ship, etc. – when you have that technology in your hands, which I think the industry will, that’s disruptive.
McHALE: Is it all going to be in the same footprint?
BRAEGELMANN: We’re talking packages, so you’re talking [an] integrated circuit, essentially, that has all that in it. It can’t look exactly like the packages we have today, but I don’t think it’ll be dramatically different either. You’re going to get a lot into existing package architectures.
Jake Braegelmann currently serves as the Vice President of Business Development at New Wave Design and Verification (New Wave), where his blend of technical depth and customer-focused strategy has been instrumental in delivering innovative solutions to complex system challenges. Jake has been with the company since its inception in 2013 and has played a pivotal role in its growth, drawing on his extensive industry experience including as an engineer within Lockheed Martin’s tactical airborne processor group. He holds a master’s degree in computer engineering with a specialization in embedded design.
New Wave Design • https://newwavedesign.com/
