Tomorrow’s battlefield will use local, mesh, and joint service sensors, fusing them together into a common operating picture that provides extreme situational awareness, improves lethality, shortens the sensor-to-shooter decision loop, and uses fewer human resources to increase battlefield success.

Deployed and edge processors, with advanced in-vehicle high-bandwidth LAN networks with switches, routers, and red/black domains are needed to interface with these sensor arrays.

 

GMS builds several available-right-now rugged COTS products that can meet these needs.
This Application Brief maps keywords and battlefield needs to various GMS products.

Multiple Intelligence (Multi-INT) processing:

The small form-factor (SFF) GMS S422-SW or S422/X422 combination (server/dual GPGPU co-processor) are ideal deployed or edge servers with 30 10GbE ports. An upgraded roadmap product is the S522 dual scalable Xeon CPU rugged server for multi-sensor aggregation and network transport.

High-rate sensors like RADAR, LIDAR, EO/IR and other camera/video imagery can be moved on 10Gbps links and aggregated into 28 ports in near real-time with minimal latency. Dual 100 Gbps up/down links and intra-chassis connections link multi-processors or provide reach-back. These systems are “shoebox”-sized small form factor, conduction-cooled ultra-rugged chassis designed for passive cooling in armored, airborne, and unaided platforms. High-bandwidth, civilian cloud-based data center LAN technology is supplemented with Intel server-class processing. Servers come in single- and dual-CPU options, with Xeon E5 and Scalable Xeon enterprise processing in a battlefield package. The GMS X422 co-processor engine brings enterprise-class data mining and deep learning GPGPU processing onto the battlefield.

For the X422, add-in COTS Nvidia GPGPU co-processors are absolute leading-edge artificial intelligence engines, including Ampere-class and proven Tesla-class for FFT/IFFT, DCT, FIR and both RADAR algorithms and data mining. GMS uniquely adapts leading-edge AI engines from Nvidia to rugged applications and assures reliable operation over the widest possible temperature range using proprietary and patented active and passive cooling technologies.

Deployable Rackmount Servers for Vehicles, TOCs and Command Posts

In addition to small form factor, rugged servers, switches and sensor processors, conduction-cooled 19” rackmount servers are also available. While less rugged than a SFF chassis, these platforms allow the widest range of add-in cards and technologies—bringing the very latest laboratory technologies right onto the battlefield for SIL-like testing or rapid deployment. GMS rackmount servers and LAN switches are convection- or conduction-cooled, have passive or active cooling, are USA-designed and -manufactured, and offer MIL-SPEC capabilities never found in commodity, commercial rackmount servers.

Key capabilities for on-battlefield, deployed, and edge processing and connectivity:

• GPGPU
• Intel server-class processing (Xeon, Scalable Xeon, second-gen Scalable Xeon)
• PCIe Gen 3 and Gen 4 interconnects
• Small form factor, conduction-cooled
• Artificial intelligence
• Nvidia with CUDA cores, vector libraries and VSIPL
• Sensor-to-LAN
• 1, 10, 100 Gbps data throughput
• Autonomous systems with high-rate multi-spectral sensors
• In-vehicle/on-platform networks, up to 28 nodes, partitioned for dual rings, self-healing, and HEU/TEU applications