Industry pushing for AI, big data to drive strategic defense decision-making
StoryJuly 26, 2023
As the defense industry grapples with exponential increases in mission data that must be collected, managed, and analyzed, the promise of artificial intelligence (AI) to harness the power of big data and drive strategic decision-making represents an unprecedented shift in the industry.
It’s not just the commercial world that is embracing artificial intelligence to manage big data. Defense industry players from commercial off-the-shelf (COTS) hardware and software suppliers to prime contractors to the U.S. Department of Defense (DoD) are integrating (AI) solutions to manage huge amounts of data that are only increasing in complexity.
Now, the question is how to use that data to help the warfighter. Industry insiders say the combination of AI and big data is poised to revolutionize defense operations by speeding up the kill chain – that necessary process of “find, fix, track, target, engage, and assess.”
Real-time battlefield applications
One major advancement for AI and big data has been its use in enabling more advanced capabilities for military personnel, particularly in real-time battlefield contexts, says Glenn Kurowski, chief technology officer for CACI (Reston, Virginia).
"A big trend is delivering increased data utilization at the edge for the warfighter," Kurowski says. “A specific example is the aggregation of multi-intelligence sensor data into a fusion framework to provide sophisticated and timely situational awareness. That situational awareness then feeds into mission planning and multidomain operations, with a mix of automation and humans on/in the loop.”
That process is easier said than done, and the solution is not necessarily more data. One aspect that the industry has focused on is the fact that the key to data exploitation on the battlefield lies not in raw processing power, but in the proper application of the data warfighters already have, Kurowski says.
Then there’s the major issue of data security: Kurowski highlights the importance of zero-trust architectures, cloud stacks, and accompanying security encryption, from data transport to live data in data lakes to storage mechanisms. One significant innovation in this area, according to Kurowski, is the easy, National Security Agency (NSA)-approved access to classified data from unclassified networks, enhancing data access for users at the edge via Commercial Solutions for Classified Programs (CSfC).
"It now includes over-the-air rekeying, integrated retransmission devices, and a more turnkey accreditation process," he says. "Essentially, there’s now an easy button for one of the most pressing operational needs to access data at rest from a different security domain." (Figure 1.)
[Figure 1 | The CACI CSfC solutions are customizable, scalable, and National Security Agency (NSA)-compliant for maximum edge security. Stock image.]
Integration is an ongoing obstacle
Transferring data across domains is a key driver in enabling the military's Joint All Domain Command and Control (JADC2) effort, but the reality of this integration is harder than it appears.
To address the difficulty of transferring data across older systems, Kurowski advocates for two specific strategies: optical communications and rethinking data utilization.
Optical communications “requires highly secure, high-bandwidth, resilient paths of network communications using photonic energy (lasers),” he adds.
The second strategy requires changing the approach to data utilization. “Think beyond ‘data’ being contained at a ‘location’ but rather to accepting that it exists everywhere and focus on managing accessibility and having resilient net-works to ensure access,” Kurowski says.
Stephen Carlon, managing director and client account lead for C4ISR at Accenture Federal Services (Arlington, Virginia), says his firm attempts to tackle this challenge using technology called PICARD (Platform for Integrated C2 and Responsive Defense), which is aimed at assisting with the mixture of old and new systems.
“By normalizing the data into common formats, it removes the barrier of requiring systems to conform to a certain data standard,” Carlon says. (Figure 2.)
[Figure 2 | A visualization enabled by Accenture's PICARD from NAVCENT's Digital Horizon exercise in December 2022.]
OSD getting involved
The effort to figure out how to best utilize big data is not just an industry focus – the Office of the Secretary of Defense (OSD) has also made it a priority. The OSD stood up the Chief Digital and Artificial Intelligence Office (CDAO) in June 2022 in an effort to centralize OSD’s digital and AI capabilities to ensure unity in the implementation of AI and data in general, says Lt. Cmdr. Tim Gorman, OSD spokesperson.
CDAO's initiatives are focused on enabling self-service to easily access data, receiving input from all data users through feedback loops, and measuring value with metrics to support growth and quality.
CDAO also hosts the API [application programming interfaces] marketplace, a portal powered by Google Apigee (Google’s API platform), Gorman adds.
"Users discover data products through a self-service developer portal,” he says. “The data product is designed to meet the specific business and mission needs of user groups. These products are understandable, high-quality, secure, easy to access, and made available through a well-defined interface.”
To fulfill the goals of becoming a more "datacentric DoD," CDAO has been developing an ontology – a collection of terms describing the types of objects and/or events and their relationships that comprise a domain of interest. Implementing a defense-wide semantic layer enables the linking of enterprise data, Gorman says.
The goal is to integrate operational and intelligence data and analytics to provide commanders with a greater understanding of the battlefield and a decision advantage over potential adversaries.
The impact of MOSA and interoperability standards
In order for both industry and DoD officials to be successful in leveraging big data successfully, it’s increasingly clear that open standards and interoperability will be key – especially with the push toward more cross-domain access to that data.
Carlon points to initiatives such as the Integrated Sensor Architecture (ISA) currently in use by the Army, which enables “the rapid acquisition and deployment of new data in a mission-forward, sensing environment," he says.
The continued refinement of the modular open systems approach (MOSA) in the defense industry, which includes the adoption of standards for data encapsulation and formats, has been vital, Kurowski says: "There are working groups defining standards for data formats, descriptions, and storage."
For security, the focus is on zero-trust architecture. Kurowski also pointed to the Common Modular Open Suite of Standards (CMOSS) as an example of defense-industry standards integrated into program requirements.
Carlon also sees the value of MOSA concepts for data interoperability. He points to Accenture's own data mesh solution designed to acquire data from unlimited sources and deliver it to numerous data consumers. "By implementing this two-way MOSA strategy, complex data environments become easier to manage," he says.
Tim Stewart, director of business development at Aitech Systems (Chatsworth, California), says MOSA is key because it "promotes interoperability, flexibility, and reusability."
When it comes to interoperability and security challenges, Stewart says that MOSA’s biggest industry benefits are standardized interfaces, interchangeable components, plug-and-play integration, scalability, and upgradability.
Balancing democratization and security of data
The democratization of data – or the act of making the data easily accessible to those who need it – is another internal push driving the utilization of big data in defense. However, more easily accessed data comes with its own set of challenges, primarily revolving around security concerns and system innovation.
"We need to liberate data from traditional silos," Kurowski says. "It does require us to do so smartly – with the proper security controls – but we have the technology to do that. In my opinion, this isn’t a technology problem. It’s a policy, legacy-control, and sometimes even a title authority issue. And, of course, there needs to be a robust insider threat program.”
Carlon acknowledges the tension between the need to provide data widely and the need to control access to it – and this is where innovation can help.
"As AI/ML models get better and faster with more data, that allows us to process and receive more and more data, improving our understanding of the information environment,” he says.
Ultimately, the industry will need to find a balance, Stewart says: "Access controls, data governance, and privacy must be implemented so that data is available only to authorized personnel and otherwise protected," he says. (Figure 3.)
[Figure 3 | The Aitech A179 AI supercomputer is a low-power, ultra-small-form-factor GPGPU-based platform aimed at improving performance at the edge.]
The future of big data and AI in defense
While many issues remain, those involved in the defense industry understand the opportunity that big data provides to the future warfighter and are working to figure out how best to leverage it.
"The intersection of big data and AI in the kill-chain model ultimately becomes the foundation for improved and better-informed decision-making," Kurowski says, noting that CACI's Information Advantage is focused on analytics, data optimization, AI functionality, and platforms to offer an advanced intelligence picture for operational decisions.
Carlon says AI/ML has a great deal of potential to make data more useful: "Pushing AI/ML as close to the edge (that is, where the data is first acquired) will be the key to rapidly exploiting data and speeding the kill chain in the future," he says. He adds that this strategy is particularly necessary in denied, degraded, and limited (DDL) environments where adversaries may restrict communication capabilities.
Big data can, in turn, make these AI models better, which opens up all sorts of possibilities for defense contractors, Stewart says. His company, Aitech, offers a low-power AI supercomputer designed for performance at the edge.
“Big data provides the data to train AI models. AI enables the extraction of insights at a rate that would otherwise be impractical,” he says. “This relationship expedites the processing of information, accelerates decision-making, and speeds up the kill chain by identifying patterns and threats in real time."
