NASA wants high-performance spaceflight computing capabilities

Posted by Thomas Anderson

NASA and the U.S. Air Force Research Laboratory in Albuquerque, N.M., are requesting research and development proposals to define the type of spacecraft computing needed for future missions.

Through a broad agency announcement, the Air Force Next Generation Space Processor Analysis Program is seeking two to four companies to perform a yearlong evaluation of advanced, space-based applications that would use spaceflight processors for the 2020-2030 time frame.

NASA's decision to partner with the Air Force and issue a joint solicitation was influenced by a four-month formulation study funded by NASA's Space Technology Mission Directorate's Game Changing Development Program. During that investigation, engineers from NASA's Goddard Space Flight Center in Greenbelt, Md., NASA's Jet Propulsion Laboratory in Pasadena, Calif., the Johnson Space Center in Houston, and NASA's Ames Research Center in Moffett Field, Calif., evaluated 19 real-life mission scenarios involving the use of flight processors.

"We surveyed NASA's needs and it became more than obvious that we could take advantage of an advanced processor," said Richard Doyle, the program manager for JPL's Information and Data Science Program and study leader.

By any standard, NASA's state-of-the-art is significantly less capable than what is available in most consumer products, said Wes Powell, a NASA Goddard engineer who participated in the study.

"We have special requirements," Doyle said. "Our flight needs are more extreme and our processors must be able to perform robustly in a radiation environment, using low power." As a result, both military and civilian mission planners must use specialized, vastly more expensive processors that have been hardened against radiation-induced upsets and generally have a higher degree of fault tolerance.

The current state-of-the-art — the RAD750 — is a single-board computer manufactured by BAE Systems Electronic Solutions. Specifically designed to operate in high-radiation environments like those encountered in space, BAE released the technology in 2001 as the successor to the RAD6000. As of 2010, the RAD750 had become de rigueur for a broad range of space missions, including the Curiosity rover, the Solar Dynamics Observatory and the Fermi Gamma-ray Space Telescope, among others.

Though it's hardened against radiation-induced upsets and uses only five watts of power — another important performance requirement in energy-constrained spaceflight missions — the RAD750 computes only 200 million operations per second.

To get around these computational limitations, mission designers are implementing highly customized processors featuring more powerful, radiation-hardened, field-programmable gate arrays, which are capable of implementing application-specific processing circuitry. While these custom-designed processing solutions handle heavier data loads, they can be difficult and time consuming to program, and aren't as power-efficient for general purpose processing, Powell said.

"What NASA needs is an energy-efficient general-purpose processor capable of billions of operations per second, thereby making it applicable to most missions," Powell said. "The bottom line is that while the RAD750 has been very successful, it is generations behind the current state-of-the-art."

In addition to establishing the rationale for a technology investment, the team surveyed six different architectures and decided that multi-core processing satisfied NASA's objectives. With multi-core technology, a single physical processor contains the core logic of several processors, which are packaged into a single integrated circuit. Multi-core technology is used in desktops, mobile PCs, servers and workstations, and allows the system to perform more tasks and scale its energy consumption, depending on what is needed at the time. 

This isn't to say that the effort won't face challenges. "The key challenges are processing throughput, radiation and fault tolerance, power efficiency, and the ability to broadly scale power and performance, using no more than seven watts," Powell said.

The time for change, however, is now, he said. "We need a significant increase in performance and power efficiency. A small incremental improvement won't justify the investment. The development of a spaceflight multi-core processor will provide transformational improvements in onboard processing for NASA's future missions."

The NASA – Air Force call for proposals was made in April by NASA's Space Technology Mission Directorate, which is innovating, developing, testing and flying hardware for use in future science and exploration missions.