FPGA pioneer Xilinx is targeting a range or military and space applications with its latest generation of “defense-grade’ programmable logic chips, including expanded military use of machine learning and AI applications as well as securing devices at the network edge.
Along with standard features like a ruggedize packaging and resistance to temperature extremes, Xilinx said its latest generation of military grade systems designated XQ also includes anti-counterfeiting and reliability features for systems that must often operate for decades.
The chip maker’s programmable logic devices are already integrated into front-line weapons such as the F-35 fighter, and Xilinx is stressing the security and anti-counterfeiting features of its latest FPGAs and SoCs based on its Ultrascale architecture. The “Xynq” portfolio of programmable silicon for harsh military and space applications includes new multiprocessor and RF SoCs with an emphasis on scaling and security. The chip maker is promoting those security and anti-counterfeiting features for emerging machine learning applications like autonomous systems that must process data locally while remaining hack-proof.
Those emerging applications underscore the growing need for programmability as threats and requirements evolve, especially as the Pentagon begins to leverage machine learning. Hence, Xilinx and other vendors are striving to meet requirements for common platforms that can be modified to deter new threats via software updates.
David Gamba, senior director of the aerospace and defense unit at Xilinx, also stressed the “footprint compatibility” of the Zynq XQ processors, noting that device density can be adjusted up or down depending on the application.
Xilinx is among the earliest suppliers of programmable logic for military and aerospace systems, and the Zynq XQ portfolio represents its eighth generation of military-grade devices since the San Jose chip maker was founded in 1984. Major military contractor such as Raytheon “are looking for single-source suppliers,” Gamba told EETImes. “The key message here is consistency” and the ability to upgrade existing platforms with new generations of programmable logic while maintaining a long logistics tail for older parts.
That translates into the ability to design-in Xilinx FPGAs over a four- to six-year span, with new technology nodes arriving about every two years. The latest SoCs as well as Kintex and Virtex FPGAs are based on Taiwan Semiconductor Manufacturing Co.’s 16-nm FinFET process.
The programmable logic lineup also addresses obsolescence within the defense electronic supply chain and the counterfeit chips that can result when supply chains run dry. Hence, the new Xilinx processors incorporate a new security feature called PUF, for 256-bit physical unclonable function. The processors and FPGAs also include anti-counterfeit markings that serve a unique fingerprint for each device, Gamba said.
These approaches, combined with making devices available for military and aerospace systems for up to 20 years, aim to secure the defense electronics pipeline by reducing the number of obsolete parts.
Other anti-counterfeiting features include a feature called mask set control that prevents changes to a chip design. The goal is to keep chip intellectual property from falling into the hands of adversaries, Gamba said.
Among the military applications for the new Xilinx chips are avionics, communications, electronic warfare and radar systems. For space applications, Gamba said satellite communications is the fastest growing segment.
— George Leopold is the former executive editor of EE Times and the author of Calculated Risk: The Supersonic Life and Times of Gus Grissom (Purdue University Press, Updated, 2018).