Protecting and Analyzing Data from Unmanned Platforms at the Edge of the Battlefield

The future of combat will increasingly leverage both unmanned and optionally unmanned fighting vehicles. A key driver for the use of unmanned platforms is their ability to collect data without putting warfighters in harm’s way.  A resulting challenge is how to store the massive amounts of sensor data collected onboard the unmanned platform and how to ensure that data is secure in case the platform should get lost and fall into the wrong hands.Adding to this problem is the inexorable rate of increase in sensor data resolution. For example, where we had HD video just a few years ago, we are now seeing requirements to support 4K video, which will itself soon be supplanted by 8K video.

To turn all of this valuable data into actionable intelligence requires significant amounts of processing, some of which can take place onboard the platform if powerful and rugged enough compute resources are available. Onboard processing enables a reduction in the size of the data, enabling key data to be downloaded in real-time to analysts at the Forward Operating Base (FOB). Unfortunately, the data downlink transports available from unmanned vehicleshave been unable to keep pace with the firehose of data that these platforms are now able to collect and store.

The good news is that advances in processing technologies, such as the use of GPU enabled devices to drive AI and ML applications, can help optimize data sizes to fit through a real-time pipe. Using a “store and forward” approach, a rugged high density storage system onboard the platform can be used to store and protect all of the collected data at full resolution for post-mission analysis, after the drone, for example, returns to base. During the mission, subsets of sensor data (think of low-resolution thumbnail images) can be created by compressing the data or adjusting the sampling rateto produce an acceptable and “good enough” representation of that data to fit into the real-time transport pipe for immediate transmission. This approach can quickly provide usable information, such as sensor, video, positional, thermal, or fuel data, for example.What’s more, forwarding thin-pipe level data in real-time speeds the process of identifying what data is most important for analyzing once thefull resolution videoand datastored on the platform returns to the FOB, where it can be reviewed post-mission. similar to how an NFL football coach can study in detail the recording of Sunday’s game on Monday.

During some unmanned missions, probability of intercept/probability of detection concerns will make the real-time downlinking of data untenable. Once a vehicle is sent out into the battlefield there’s a greater chance of the vehicle and the critical data it has collected falling into adversarial hands. For those reasons the sensor data must be stored using a system with Data-at-Rest (DAR) protection. By leveraging multi-layer commercial encryption,as outlined by the NSA’s Commercial Solutions for Classified (CSfC) program, critical data can be encrypted for storage and forwarding in an NSA approved manner. This provides an alternative to Type 1 encryption with greatly reduced risk and consequences in the event the platform or data is captured by the enemy. A good example of a data storage system that can protect terabytes of DAR using CSfC is Curtiss-Wright’s DTS1 Network Attached Storage solution.

Unlike airborne platforms, unmanned ground vehicles and naval platforms can also benefit from recent advances in data communication. For example, the advent of 5G networks in the battlefield can establish a large high-speed data communications “bubble” that enables collected data to be processed in the field in real-time.

Typically, the size of the vehicle and its distance from the FOB will determine whether or not sensor data transport can be performed in real-time. An example of a rugged processing system for data analysis is Curtiss-Wright’s PacStar Tactical Fusion system, which can be used to search for signal of interest in data captured from drones in real-time to the FOB, or can be used to process the full high resolution data set after removal from the returned unmanned platform. Today, sophisticated battlefield networks can be quickly stood up to connect the FOB to the Cloud. What’s more,if connectivity is denied or reduced, the Cloud can be replicated locally at the FOB for limited amounts of times. Curtiss-Wright’s PacStar MDC system is a compact, mobile system that can act as a local Cloud or as a Cloud synchronization point, so that multiple battalions in theatre can share a disconnected Cloud. This enables unmanned vehicles to continue transmitting sensor data, which is stored temporarily in the replicated Cloud, and when connection is restored, can be re-synchronized with the appropriate government Cloud, making the critical data available to leadership.

“Today, Sophisticated Battlefield Networks Can Be Quickly Stood Up To Connect The Fob To The Cloud”

If there’s one biggest technological advance making possible the recent breakthroughs in the miniaturization and ruggedization of deployable enterprise networking/computing solutions it’s the advance of modern GPU hardware, the software that leverages those GPUs, and the availability of ruggedized versions of those devices, from vendors such as NVIDIA, that can be used in deployable military systems. Additional factor helping to drive these capabilities are the increase in solid state storage density that’s appropriate for field usage, such as NVMe devices, and the inevitable march of Moore’s Law that continues to drive new levels of processing power on a regular drumbeat.

Along with the ability to collect and store sensor data in the field, the DoD’s mandate to use electronics systems based on the modular open system approach (MOSA) has helped increase the use of standardized interfaces such as Ethernet and IP networks. Where before system designers may have selected, for example, dedicated video buses like SDI, that were difficult to integrate, the use of Ethernet is becoming ubiquitous. Previously, a platform might have deployed an analog sensor that required a dedicated piece of translation equipment to convert to the data into digital format. Today, advances in electronics enables those sensors today todirectly use a common data format or use a published API. That means that the unstructured store of data, sometimes referred to as a “data lake,” can be more rapidly exploited using powerful analytic software, such as Kinetica, which can process the unstructured data at a rate not even imagined a few years ago.

All of these advances combine to empower the warfighter and decision makers with more data, which helps them observe, orient, decide, and act on the most informed path.