Sponsored Story

Extending Human Performance Through Technology: The Promise of JADC2

The Air Force’s vision for the seamless, networked warfighting concept known as Joint All-Domain Command and Control, or JADC2, is all about using information to dominate the battlespace: By interconnecting systems across every domain, U.S. forces would seek to increase the operational decision space for all U.S. commanders, overwhelming adversary commanders and shortening decision-making cycles for any opponent that must react.

Information dominance is not just about connecting sensors to shooters. It also means using those sensors to create pervasive, all-domain situational awareness and to use that to cut through the fog of war.

“The part [of JADC2] that I think is going to be so incredibly game-changing is the ability for us to really use predictive analysis and inform our decisions going into the future,” said Gen. Terrence O’Shaughnessy, the Air Force general who was formerly in charge of U.S. Northern Command, in a video conversation with the dean of the Mitchell Institute for Aerospace Studies, retired USAF Lt. Gen. David Deptula. JADC2 will provide decision advantage, he said, helping commanders “make decisions that, like playing chess, are thinking about two or three moves downstream…It’s going to give the decision-makers—at the speed of relevance—the ability to make really complex decisions.”

Video: Mitchell Institute on YouTube

However, achieving the speed of relevance requires changes and means thinking anew about how platforms are built and sustained.

Dr. Bill Conley, former director of electronic warfare at the Pentagon and now chief technology officer at Mercury Systems, says the conventional model locks in requirements early and takes years to develop. By the time systems are fielded, the once cutting-edge technology inside is obsolete. By contrast, commercial technologies and standards are continuously evolving. The Pentagon, he says, needs to leverage that speed.

“We need continuous improvement, continuous delivery, and continuous development of capabilities,” he says. “Having the ability to continuously integrate state-of-the-art commercially available technical solutions into evolving systems is the only way to keep up with the speed of change in the technologies JADC2 will employ.”

Sentry at sunset
Maintenance Airmen prepare an E-3 Sentry airborne warning and control systems aircraft for takeoff while an F-15C Eagle flies by Jan. 5, 2012. Photo: Tech. Sgt. Arian Nead

Enabling JADC2

Whether deployed on the ground or in the air, evolving technologies cannot be limited to command centers. Military personnel at the tactical edge need both high-end computing and connectivity to enable JADC2, Conley says.

“Air Force AWACS aircraft operate like orchestra conductors,” Conley says. “They assign targets to different fighters, and those fighters go out and execute their assigned mission. But in this model, only the AWACS is responsible for understanding the big picture.”

That won’t fly in an era of great power competition. In the future, peer competitors who spent the past 20 years watching the U.S. military at war will seek to break down U.S. strategic advantages in command and control, communications and intelligence gathering. In response, the U.S. must up its game.

It’s not enough to have one centralized command center making all the decisions; if it fails, there’s no recourse. Rather, the entire network must be resilient enough to overcome such a loss, with every aircraft in the operation able to operate both singly and as part of a team.

“The enemy always gets a vote,” Conley points out. “Part of this is thinking about the problem locally — I have to make a good decision with the information I have.”

Modern warplanes are bristling with sensors, but they also need the ability to process that data and use it to help pilots make decisions. Pilots are already operating under information overload. What they need isn’t more information, but better information, made more accessible through intelligent display.

Two technologies are critical to making human performance a force multiplier, rather than a bottleneck, Conley says:

  • Visualizations and displays to communicate data to the pilot in more intuitive, digestible ways.
  • Artificial intelligence, to contextualize and help interpret data so systems offer options and solutions, rather than data to be analyzed.
Pilots at Eglin Air Force Base, Fla., use Full Mission Simulators as part of their training with F-35s. The F-35 Full Mission Simulator accurately replicates all sensors and weapons to provide a realistic mission rehearsal and training environment. Photo: Lockheed Martin/ courtesy, via DOD

Advanced Displays for Advanced Missions

Displays must be flexible and reconfigurable so a pilot can get information in the best possible form depending on where they are in each mission. “As you’re taking off, you care about a different set of things than once you’re in level flight going toward a target, and also versus when you are engaging that target,” Conley said. “How do we reconfigure a display? How do we optimize the display of information to convey awareness to a pilot? And how do we provide the information to ensure pilots have what they need at each point in time?”

Just as important to Conley as reconfigurable displays is the ability to refresh technology without having to reset every training and sustainment regimen. “We need to get comfortable with the concept that we want to drop in a new generation of display technology that allows a human to fully interact with that system,” Conley says. “And we need to be able to do that on a timescale measured in years not decades. That requires an evolution in how we think about the defense acquisition system.”

RQ-4 Global Hawk's first take off from Yokota
Capt. Mark, 69th Reconnaissance Group Detachment 1, performs preflight checks on an RQ-4 Global Hawk for its first take off from Yokota Air Base, Japan, on May 5, 2017. Photo: Airman 1st Class Donald Hudson

Artificial Intelligence

Better displays and increased sensor data will be accompanied by intelligent systems that can help pilots make better decisions more quickly. But proving those systems work won’t be instantaneous, Conley says. For artificial intelligence to take hold in practice, pilots have to trust the AI will help complete the mission.

Just as it took time for drivers to become confident that algorithms could help them navigate the highways, pilots will need to learn to trust whatever algorithmic decision-making enters their cockpits. Pilots won’t surrender their judgment unless they can see with confidence that the system is trustworthy. That might mean providing confidence estimates when threats are identified, for example, or in determining if another aircraft is a friend or foe.

Figuring out how to reflect both that broad confidence level and that kind of ambiguous context in a way that users can rapidly absorb will be key to the JADC2 concept. And it’s something the Pentagon’s Joint Artificial Intelligence Center (JAIC) started working on this year, according to the Pentagon’s Chief Information Officer Dana Deasy.

The “operations cognitive assistant capability” JAIC is developing will use predictive analytics and optimized human-machine interaction “to drive faster and more efficient decision-making,” he told a media roundtable at the end of July.

“Right now, it’s all about, how do you take streams of information in and allow the machine and human to interact together to make better decisions?” Deasy said.

Incorporating AI capabilities into JADC2 technologies will pave the way for the teaming of manned and unmanned systems, as envisioned with the so-called loyal wingman concept. The autonomous wingman is flying itself, based only on the high-level goals communicated by the manned aircraft’s pilot.

High-throughput, low-latency localized data links make this kind of teaming possible. The links form a local network that shuttles command and control data between the human pilot and the autonomous sidekick.

But autonomy is the key, says Conley. “Today, we send pilots into air-to-air combat with live wingmen,” he explained. “Both of you have been trained so you know how you work together. But with manned-unmanned teaming, we have to do that without becoming so scripted that it is predictable, or it won’t really work—every adversary will figure it out.”  Conley continued on this thought, “I have to deny an adversary the ability to understand what I’m doing and why—because if the enemy doesn’t know why I did something, they can’t build a model. They can’t anticipate what I’m going to do.” 

This is the driving reason for developing systems that can be rapidly upgraded over time. According to data from Open.ai, machine-learning capability is doubling every three and a half months. So a state-of-the-art solution developed in January will be effectively obsolete one year later—only 10 percent as effective as a system rolling out the following year. Such an extraordinary pace of development demands agile development and rapid deployment of new technology into existing systems, Conley says.

“You need the ability to upgrade those systems during the conflict, while it’s deployed at a forward operating base,” Conley says. Such concepts are anathema to conventional thinking – but fully in keeping with themes advanced by current Air Force leadership, including Chief of Staff Gen. Charles Q. Brown Jr., who released his “Accelerate Change or Lose” challenge shortly after becoming chief, and Air Force Assistant Secretary for Acquisition and Sustainment Dr. Will Roper, a driving force behind the Air Force’s modernization push.  

“We need an evolution in how we think about logistics, how we think about sustaining a weapon system,” Conley says. “For AI and machine-learning technologies, we need to start thinking about upgrades as an element of routine maintenance — something that has to be done while you’re deployed—not after your deployment is over and you’re back at your home base.”

The Smart Future

Perhaps the most profound changes that JADC2 and AI will bring are to human performance, Conley said. With greater intelligence in the systems comes an opportunity to personalize the relationship between human and machine. The smarter the systems, the less generic they need to be.

Instead of a one-size-fits-all approach to human-machine interaction, intelligent systems can learn user preferences. “We should not be surprised that different pilots, different operators, consume data in different ways, just as different people learn in different ways,” Conley said. Indeed, the Air Force is already changing pilot training to reflect those different learning styles. Smart systems could allow that personalized approach to extend into the mission space.

“We can optimize performance if we can configure interfaces to best suit each operator,” Conley predicts. The key to that capability, though, is to enable more rapid technology refreshes, built around open standards that embrace the best available commercial technologies, hardened, protected, and secured to meet the most demanding military requirements. “This is why I’m at Mercury,” Conley said. “It’s what this company is about.