AUSA 2019: General Micro Systems Unveils New, Tactical-level AI Servers
WASHINGTON, D.C., October 16, 2019 – General Micro Systems is focusing its efforts on pushing artificial intelligence (AI) processing capabilities down to the tactical level, developing a family of AI-enhanced servers that are configured to withstand the rigors of combat.
Known as "edge processing", the need for AI-capable systems designed to handle information collected from "high rate, high data, high bandwidth sensors ... that send massive amounts of high resolution data" while being able to break that information into actionable intelligence from a forward location is increasingly in demand from US armed forces and allied militaries, General Micro Systems chief technology officer Chris Ciufo said.
"The intention is ... to be at the edge of the battlefield, in the vehicles", collecting raw data from airborne or ground-based sensor platforms, utilising the company's AI-equipped platform-mounted servers that meet all the size, weight, and power (SWaP) requirements of the vehicle or aircraft it is bolted on to, Ciufo told Jane's on 14 October.
The capability to drive large amounts of collected data from sensors down to servers, the server’s capability to process that information through tactical battlefield network, and the capability of the AI-driven server engines to “cull through these gigabytes and terabytes of data and create actionable intelligence from that data” as close to the battlefield as possible are the demands signal company officials are attempting to meet, Ciufo said during the Association of the United States Army 2019 (AUSA 2019) conference in Washington, DC.
S1202-XVE is the smallest variant of the tactical-level AI servers being offered by General Micro Systems. Approximately the size of a cigarette carton, the mission processor server can drive three video displays output, operates two AI processing engines, and is SWaP optimised for deployment aboard autonomous or manned fixed-wing and rotary-wing aircraft, Ciufo said.
The vehicle-mounted variant of the company’s AI tactical servers, the S422-SW Thunder and X422 Lightning, are designed to be used in tandem to provide 400 teraflops of processing speed, on par with General Micro Systems’ 1U/2U Rackmount Titan servers, Ciufo said. In comparison, the S1202-XVE is capable of processing at 10 teraflops, he added.
“Size, weight, and power mandates certain trade offs,” Ciufo said, regarding the amount of processing power company engineers were able to fit into the S1202-XVE, as well as the Thunder and Lightning AI-equipped servers, while keeping it a tactical-level system.
“You only want to downlink actionable intelligence, you do not want a [big] data stream ... because you have limited bandwidth” at the battlefield level, he said. By integrating powerful AI capabilities into an expeditionary server packaged, operators could sift through raw data quicker and make sure what is transmitted down to ground units is the intelligence they need, he added.
Read more on how GMS is harnessing AI capabilities into smaller, more survivable hardware platforms, here.
October 16, 2019