Intelligent Systems Boost End-to-End Military Operations (Sponsored Content)

May 1, 2019
By Henry S. Kenyon

Cloud and machine learning capabilities give warfighters more flexibility in the field.

As conflicts become faster and more complex with multiple platforms and data streams feeding information to warfighters, there is a growing need to manage this process to improve operational efficiency. The Department of Defense (DOD) is investing in cloud and machine learning tools and systems to help improve situational awareness and connectivity at the last tactical mile.

The military is striving to maintain tactical dominance in two ways: (1) ubiquitous edge computing and processing on intelligence, surveillance and reconnaissance (ISR) platforms, and (2) cloud-based services and tools that can reach small tactical units to provide them with vital information even in contested electromagnetic environments.

DOD ISR analysis has traditionally been a manpower-intensive process with scores of analysts monitoring live data feeds transmitted by a variety of airborne and ground-based sensors.
In recent years, the military has sought to fuse data at the edge, essentially using distributed platforms to do some of the data processing before transmitting the information to warfighters.

Not only does this approach save precious bandwidth, but by combining fused data with artificial intelligence (AI) and machine learning tools, personnel once needed for monitoring video and sensor feeds can be freed up for higher level mission-focused work, says Vernon Weisenburg, principal program manager with Microsoft Azure Government Engineering.

Simply looking at airborne platforms and focusing on video streaming misses a big part of the overall fusion of intelligence to create a comprehensive battlefield picture, notes Weisenburg. He adds that the ability to process and gain context and intelligence is not just through video, but also via audio, multispectral (RF and infrared) and other methods such as seismic ground sensors.

One of the challenges of this collected data is that there isn’t enough bandwidth to pull it all back to a tactical operations center or a command post for analysis in real time. Moving more data processing capabilities to the edge gives operators more context to work with because they are not inundated with raw data, he says.

AI and Cloud Access

While the DOD has considered the concept of edge computing/processing for some time, a convergence of technology trends and developments is now making this a viable option. This includes the commoditization of AI machine learning technologies and innovation in hybrid cloud capabilities.

But it’s important to note that these solutions at the tactical edge may have intermittent connectivity with larger networks. This requires a rethinking of cloud access and storage, says Derek Strausbaugh, chief digital officer for DOD at Microsoft. Architecting for intermittent connectivity becomes important for units operating at the tactical edge. “We presume that the customer in that tactical last mile has to be able to operate without comms, or certainly disadvantaged comms,” he explains.

Regardless of whether warfighters have connectivity back to the DOD cloud or they are working on a small form factor device at the tactical edge, the military requires the delivery of cloud access and services regardless of where they are, he explains. This requirement underscores the importance of cloud services, such as Microsoft Azure, that are ubiquitous and consistent across devices, delivering rapid access to information closer to a warfighter at the edge.

One of the technological changes permitting this is the military’s relentless miniaturization of devices such as radios, computers, servers and IT equipment from things that once filled entire data centers to single vehicle or man-portable units. This mobile equipment is capable of using AI and machine learning systems to interpret and parse incoming data flows to help warfighters get a better picture of their battlefield environment in a timely manner rather than relying on an analyst sitting in an office in CONUS, Strausbaugh says.

This is where machine learning becomes important for ISR systems such as drone sensors, Strausbaugh explains. By processing data at the edge, forward troops can use smaller, more portable systems that can analyze data that has been predigested and parsed by the sensor platform itself. Machine learning systems help make front line troops’ work more efficient and less overwhelming by locating and pointing out any anomalies or areas of interest.

These systems can run while disconnected from major networks, either by themselves or within their own small local clouds until the electromagnetic environment permits them to reconnect to larger theater and strategic networks.

Interoperability for Hybrid and Multi-Cloud

One of the key features of the DOD’s cloud efforts is a hybrid cloud design. Strausbaugh defines this as interoperability across public, private and third-party clouds. This can cover everything from large Microsoft Azure-based enterprise clouds down to a single form factor in a soldier’s hands. He notes that the goal isn’t just to move data workloads into a large cloud environment, but the agility to interoperate and move data between different types of environments. “We view it as the ability to move workloads between on-premises and off-premises. Between good comms and disadvantaged or no comms,” he says.

“The idea is that the cloud extends and becomes a full participant in your entire range of mission capabilities,” Weisenburg adds. “Cloud isn’t just a back-end capability, but it meets you where you are.”

To more efficiently manage data and bandwidth across tactical environments, the DOD has focused on edge computing and Internet of Things (IoT) capabilities such as sensors that can analyze data locally before transmitting it to warfighters. Microsoft’s Azure IoT Edge capability is a useful tool in this environment because it can run on a variety of operating systems such as Linux or Windows and across platforms. This allows almost any device to become a data source, Strausbaugh says.

“The vision is that virtually any sensor can become an IoT device,” he explains. “And it can do that over a closed-loop network provided there is some level of connectivity.”

For closed-loop tactical clouds supporting warfighters in environments where they are often cut off from larger networks, Microsoft is emphasizing device-agnostic systems. Weisenburg describes these as residing on and connecting small form factor, ruggedized devices that can be deployed during operations.

These capabilities, supported by machine learning and AI tools, help maintain the network and allow troops to make the best use of the systems in their operational environments. “We’re working to meet the warfighters where they are,” Weisenburg says. “For example, if you’re in a FOB [forward operating base] with 35 operators and one intelligence analyst and you’ve got a bunch of ISR feeds coming in and working over unreliable satcom, we want you to be able to make sense of your data, your feed and your information at the edge as much as you possibly can without having to back process it somewhere else,” he says.

Machine Learning Fights Data Overload

A major challenge that emerged with the growing number of sensor feeds since the turn of the century has been the “firehose effect” where analysts and warfighters are overwhelmed by a constant stream of raw data. One of the major advantages of AI and machine learning is that smart devices can be programmed to only provide human operators with certain types of predetermined data.

Weisenburg cites the example of using AI models to help analysts with tipping and cueing of information. If an analyst is watching a sensor feed and they are trying to count the number of objects being moved in and out of a building for hours at a time, this is where automation helps, he said. Such systems help alleviate user fatigue and can make their work more efficient.

AI models can be used to alert analysts to look more closely at something. “It’s able to provide real time potential awareness of hot spots and things that the analyst should be focusing their attention on,” Weisenburg says.

Machine learning tools can help analysts “focus on the signal rather than the noise,” Strausbaugh adds. He notes that AI tools can help tie different sensor feeds together to put together a story for the analyst to interpret. These tools can also provide alerts to direct users’ attention to something they may need to focus on and they can take on some of the more labor-intensive parts of an analyst’s job such as manually sifting through log files or monitoring hours of video feeds.

Besides helping warfighters and commanders make sense of their operational environment, AI-based tools operating in a hybrid cloud can help with overall operations. Weisenburg cites the example of logistics where smart systems on vehicles could automatically arrange for refueling or resupply at the most advantageous times and locations of their journey. Such arrangements could be done primarily via machine-to-machine communications with humans providing approvals.

To learn more about Microsoft Azure Government, visit

Enjoyed this article? SUBSCRIBE NOW to keep the content flowing.


Share Your Thoughts: