Army’s hyperconnected visions depend on new approaches to software, networks
Empowering A-teams and Pacific allies requires overcoming barriers to secure access to data.
The U.S. Army has big visions for its hyperconnected forces of the future.
Special Forces A-teams might one day include robot specialists, while other operators drop behind enemy lines to find and disable missiles long before they are launched. AI tools will enable the Army to run multiple intricate, interconnected operations around the globe, moving faster than any adversary can keep up.
But Army leaders at this week’s Association of the U.S. Army conference in Washington, D.C., said that implementing these visions will depend on ongoing efforts to change how the service buys and builds its software and secure networks.
The service aims to shift its software procurement from a highly specified “industrial-base model” to “more generic statements of need,” Army Undersecretary Gabe Camarillo said at the AUSA conference. Camarillo also said the service might use outside experts to hone its software orders.
“The reality is, there are way too many programs or contracts that had been led without an understanding of what we're asking for in a real sophisticated way. So we're looking at bringing in, for example, a team of experts that would red-team or peer-review many of the RFPs.”
Networks also need an overhaul, Camarillo said.
“Part of it is going to be ensuring that, for example, we collapse our...variety of different networks that were kluged together, into something that's more unified and more coherent,” he said.
That sounds simple enough in terms of reducing—or simplifying—the amount of code, computers, personnel, and IT practices within the institution of the U.S Army. But the service’s Pacific plans are bound to bring in a lot more of the above as they tighten training, operational, and data connections with partners with their own legacy IT: the Philippines, Indonesia, Malaysia, and others.
But if data can be secured independent of the network it travels on, that could enable the Army to use a much wider variety of systems, from commercial networks to military gear.
“We're focused right now on securing the data, not just the pipe,” said Maj. Gen. Chris Eubank, who leads the Army’s Network Enterprise Technology Command.
Ultimately, said Brig. Gen. Kevin F. Meisler, head of the Hawaii-based 311th Signal Command, operators near China will be able to rely on “all forms of communication, everything from SATCOM down to [high frequency], and everything in between. And we have to be very resilient or redundant to make sure that if something does happen in one area, we can quickly switch to another.”
A lesson from Ukraine is that it’s risky to depend too much on a single company for satellite communications. Meisler said the Army is making sure that that isn’t a concern in the Pacific.
“We actually rely on multiple partners or commercial partners for all four [orbits]”: geosynchronous, medium-Earth, low-Earth, and highly elliptical, he said.
The success of Elon Musk’s Starlink is drawing others into the market, Meisler said.
“The competitors are starting to move forward and understand, you know, how to get something faster, quicker, at a reasonable rate. And we are investing in multiple different venues, and avenues to ensure enough diversity,” he said.
The goal is to build out networks, such as the Mission Partner Environment in the Pacific, that allow partner militaries like the Philippines to quickly access data that isn’t classified but that could be important or time-sensitive.
Officials described how connectivity, whether provided by military or commercial networks, is providing new opportunities for the Army to be in many more places at once. That’s especially important now as the service contends with growing capabilities from China and increasing belligerence from Russia, among other crises such as a massive recruiting shortfall.
Linking special operations forces with new intelligence feeds via satellite should help the Army better understand and thwart new missile threats, said Lt. Gen. Dan Karbler, who leads Army Space and Missile Defense Command.
Karbler said his command is experimenting with a “missile-defeat-effects coordinator” who would direct efforts to “disrupt, delay, to disintegrate adversary missile capability left of launch, so whether that's in the silo, still on the…runway, somewhere before that missile, that aircraft, has taken off.”
U.S. special operations forces will increasingly use space and cyber capabilities to stealthily disrupt enemy plans, said Lt. Gen. John Braga, who leads Army Special Operations Command. That looks like “asymmetric, non-attributable options, flexible term, collective response options for the joint force. That's what we're trying to do as an experiment here,” to get at a variety of enemies “Whether it's a cyber network, a space network, a missile network, a terrorist network.”
Braga and Karbler said that secure access to data is essential to those visions.
Data exchange at the tactical level will also be important to help soldiers achieve new capabilities, like operating robotic systems, whether on the ground, in the air, or at sea, said Brig. Gen. Guillaume Beaurpere, commanding general of the U.S. Army’s John F. Kennedy Special Warfare Center and School.
Beaurpere said another lesson out of Ukraine was that allies and partners achieve much better results when they have a great training relationship with their U.S. counterparts. In the future, that could mean developing a new speciality, particularly in special operations, for robot operation–and more importantly–training partners to do the same.
“That means training a partner force in a robotic capability. How do you integrate robotics into a scheme of maneuver for ground combat? I think that's a critical requirement of the future. And that's something that we're exploring right now with our operators. How do you give them that skill? Is it a new military occupational specialty, or is it an additional skill that is acquired by an operator that has baseline skills that is a master of the basics, but then has additional skills with robotics and then that goes forward into theater at the edge with a partner.”
Whatever that data network looks like, said Beaurpere, elite operators shouldn’t be overloaded with too much information that they can’t use.
Eubank said keeping that information flow tight, focused and relevant is a key concern. “From a networking perspective, we're trying to pull complexity as far up as possible so that [operators near combat] don't have to worry about complex tasks.”
But getting all the information that the military collects to seamlessly flow where it needs to be across service and combatant commands is some ways off, said Margie Palmieri, deputy chief digital and AI officer at the Chief Digital and Artificial Intelligence Office, or CDAO.
“Does the information actually flow in the way we expect it? A lot of times, actually, I'll say in the three experiments we've done so far this year, we've always found something that didn't quite work the way we expected,” Palmieri said, citing security concerns, configuration, and bureaucracy.
Often different portions of the military and different services label things differently, which can prevent machine learning programs from finding the right data and getting it where it needs to go, which is the why the office is now heavily focused on discovering ontological relationships between different items, whether it's what a weapon is called, or a photo, or something else across different services.
Another big goal of the CDAO office is to help the military prioritize the communication of fallibility in data, a simple principle of Bayesian inference, which lies at the heart of machine learning: understanding the probability of being right (sometimes called a confidence interval) is critical information for people working with artificial intelligence, especially in the military.
“We know that the environment of warfare is going to change as soon as we come into conflict,” she said. “And so what we've trained an AI algorithm on in the past is not necessarily what it’s going to see going forward. A tank looks like different whole versus blown up. And so the confidence levels will change as the algorithms interact with their environment. And our users have to understand that.”