DARPA's Director of the Strategic Technology Office wants the US military to rethink R&D and adopt lessons learned from Silicon Valley.
Picture a tank. It's perhaps the most iconic image representing the military: a tank is a large monolithic system built of complex interdependent parts. It's sturdy because that's exactly what it was designed to be: years of research and testing went into each component, from the track to the armored skirt to the radar, before research even began on how to integrate those components into a ground vehicle.
While this monolithic approach to system design served the US military well for decades, Dr. Timothy Grayson, Director of the Strategic Technology Office at DARPA, wants to rethink the way they build warfighting capabilities.
He's advocating for a "system of systems" approach that he's calling mosaic warfare: the ability to piece together different systems to build new overarching warfighting capabilities. Grayson says, "We're asking, what are the technical solutions that will enable faster uptake of new capabilities and the ability to field a new war-fighting capability from combinations of existing elements?"
Not only does Grayson want the US Department of Defense (DOD) to shift to making and operating a mosaic-like system of systems, he wants the military to learn to do it blazingly fast.
One Eye on Silicon Valley
In Silicon Valley, new products are built on top of cloud computing, and they're rolling out at a mind-numbing pace. Whereas just a few years ago, software companies were boasting about releasing tens of new software products or updates per day, some large companies like Amazon now claim that one deployment takes place every second. The DOD, on the other hand, moves at a slower pace. "In the DOD, even a software-intensive program, we're lucky to get five to ten new capabilities in two to five years," says Grayson. The difference in speed is orders of magnitude.
In order to come up to speed, the DOD will need to find a way to utilize capabilities that are already available—fielded military systems or commercial—and figure out how to integrate them into the broader defense architecture. Although utilizing commercially available technology may sound like a security concern (we don't want a military system that can be hacked as easily as, say, Equifax), it's possible to use a commercially available core technology but maintain a competitive advantage through implementation or integration.
For example, Grayson points out that Amazon Web Services (AWS) is a global business, and billions of people use their core cloud technology. "We might say, ‘Hey, AWS, go build us a completely offline, not part of your commercial cloud, copy of what you've got,'" says Grayson. The version AWS builds for the DOD could have additional security features and be on a different network, giving it unique capabilities, but the core cloud service would be technically identical to the commercially available version.
However, rapid integration is not straightforward for the US military. Most military systems run on special-purpose embedded computers with much more demanding performance needs than commercial clouds, and they currently require reliability and security to be proven up front. Technology is often tested in a lab or test range for years, under every possible condition, in order to anticipate every way that something might go wrong before it's integrated into a platform. Instead, Grayson advocates for a more modular system that can be tested in real time. For example, if a new image sensor starts to act funny and the legacy platform it's attached to starts to underperform, they need to be able to scale back or cut off that new sensor while retaining the legacy capability.
Grayson acknowledges that this approach might sacrifice some function of the platform—it would be down a sensor, after all—but believes that it's a worthwhile exchange for the ability to ingest the new capability quickly—much more quickly than current practice.
In the commercial world of startups, changes and additions are small and incremental—baby steps—and as a result, the risk and complexity of adding something new is minor. This mentality is a very different than the US military's approach to building monolithic systems, which are, well, built like a tank. Grayson wants to apply the "baby steps" model to military technology, because simple and low-risk systems can be built at speed. Grayson notes, "It doesn't help to take baby steps if you are still walking slowly."
New communications, integration methods, and testing methods will be needed to fully realize the mosaic warfare concept. Artificial intelligence and other kinds of automation will also be needed so that operators can interact with new tools without months of training. Even mundane aspects of system design, like maintenance and reliability, still need to be worked out.
High-energy lasers, for example, have been of great interest to the DOD, and there have been successful experiments demonstrating their performance. But those experiments involve a small army of PhDs cleaning the optics and keeping the system aligned. "Are you going to do that on the battlefield when you want to use this thing?" asks Grayson.
While many R&D programs create great technology, it often falls into the valley of death because usability is beyond the scope of the organization that developed it. There's work to be done to make these tools practical and reliable, and Grayson wants more people in science and engineering thinking about these types of problems. That's an area where the DOD can leverage the know-how of the commercial marketplace. Private industry knows how to move a new technology from concept to a revenue-earning product, and they know how to work out issues around usability and reliability in the process.
In his plenary at SPIE Defense + Commercial Sensing, Grayson told the audience where the DOD is looking for ideas, and he's hoping this will start a two-way conversation. He says, "As you're probably aware, DARPA's currency is the innovation and ideas of the marketplace, like that at DCS. We have no labs of our own. All the good work happens out in the companies and the universities."
|Enjoy this article?
Get similar news in your inbox