JUST IN: Pentagon Needs to Organize Its Data, Official Says
The Defense Department is in the midst of a digital transformation, and arguably the most daunting piece is the combined joint all-domain command and control effort, known most recently as CJADC2. Central to the concept is a unified network, and to pull it off, the department needs to enable better data interoperability across the joint force, said one of the leaders of the effort.
That goal is a top priority for the Chief Digital and Artificial Intelligence Office, said its deputy, Margie Palmieri, speaking at the National Defense Industrial Association’s Emerging Technologies for Defense conference Aug. 30.
“We're on a path now for CJADC2 where the systems we have developed as a department have really come up in silos, traditionally,” she said. “We've had a very industrial approach to how to build something — requirements on the front end create a box that that system is supposed to operate in, and a new connection or customer [to] that system has to kind of start again and go through the requirements process to be enabled.”
Palmieri said her office has “dived into this … looking at how we can enable better data interoperability across the joint force.”
She called CJADC2 a department-wide effort that will require changes in technology concepts, processes, policy and “definitely our people and especially how they think about conceptual command [and] control.”
Specifically, within the chief digital and artificial intelligence office, there are three components focused on CJADC2, the first being data integration, Palmieri said.
The data integration layer is not a new application, she said. “It is a mechanism by which we can take the different approaches to data that exist from the strategic level, the operational, all the way down to the tactical level and our tactical systems and dealings and processing and make sure that it's interoperable.”
Data integration means a tracked or targeted object remains the same across all echelons and all commands, regardless of what application is being used, she said. “You can interpret that in the same way, and you can have the same data around what that is.”
The second piece is applications. The way people work with technology and interact with information “is key, and applications are essential to that,” she said. Developing applications in conjunction with users — and leveraging feedback and lessons learned — is part of the command-and-control approach going forward, she said.
The third piece is experimentation.
One example is the Global Information Dominance Experiment, started by U.S. Northern Command and picked up by Palmieri’s office earlier this year. The experiments are designed to test, measure, optimize and field CJADC2 solutions using a unified data layer that is vendor agnostic and supports the application of machine learning and artificial intelligence models.
Palmieri said the experimentation has proved a “great way to learn what works and what doesn’t, especially in the space of [command and control].
“Because we know that as technology evolves and concepts evolve, we’re going to have to do that together. And when they work together in experimentation we can learn faster,” she added.
The experiments also provide insight into how combatant commands and the joint staff work together to “make sure we have the right force packages going where we need them and the right strategic decisions being made around deterrence before we even get to a potential conflict,” she said.
One enabling tool across all three components is the use of artificial intelligence and a department-wide goal to decentralize how its approached. Building artificial intelligence going forward, “we don't do it in the same silos in which we did our non-joint domain-specific command [and] control of the past,” she said.
“We want to make sure we have the same level datasets across all of our AI algorithms so that we know that the AI is going to learn from the right information,” she said. Her office is also looking at test and evaluation approaches that can ensure operators can “justify competence” of the artificial intelligence data they’re receiving and can incorporate the data into their own decision making.
Palmieri called the joint effort a strategic advantage, saying the recent addition of ‘C’ to the CJADC2 acronym, for combined, is a messaging change that reflects an approach taken “for quite some time.”
“As we design capabilities, we’re finding that sharing information, sharing data with our allies and partners and ensuring interoperability as we come together as a joint combined force has to be baked into all of our solutions. And so we wanted that to be front and center,” she said.
The CDAO has partnerships across the department with the joint staff, the services, acquisition and sustainment, research and engineering and the chief information officer “to really make sure that pulling together these capabilities in a joint and combined way is something that we're pulling from all elements of [command and control] that have to come together to make that happen,” she said.
The progress made on CJADC2 hasn’t been given the attention it deserves, she noted, referencing analysis, research and development, engineering and rapid experimentation efforts across the department.
The foundation is there, she said. What is lacking is the ability to field the capabilities at scale.
“And how do we take in the individual efforts that we've [engineered] and now make it ubiquitous. And we think we can do that through how we approach data,” she said.