Ch6-1

Computers in Spaceflight: The NASA Experience
- Chapter Six -
Distributed computing on board Voyager and Galileo

[172] Voyager and Galileo are two outer planetary spacecraft that carry extensive computing capability. In spectacular encounters with Jupiter and Saturn, Voyagers 1 and 2 returned science data and imaging that far exceeded results of previous planetary flybys. Uranus was the successful 1986 objective of Voyager 2, nearly 10 years after launch. Galileo is designed for a Jupiter orbiter and probe mission.^* Both types of spacecraft carry multiple computer systems, distributing functions among several machines, rather than using one central computer system as on the Viking Orbiter and Lander.

Distributed computing on large unmanned spacecraft developed conceptually from several sources. In 1967, Marshall Space Flight Center commissioned a study by General Electric Corporation's Missile and Space Division in Philadelphia as part of preparation for a huge "Voyager" Mars lander to be launched on a Saturn V booster in the early 1970s. Marshall asked GE to compare the advantages of a central computer configuration versus separate computers for different subsystems. General Electric used a highly mathematical approach to develop power, size, and weight comparisons of the different proposals in light of reliability considerations. Computer physical limits were set as high as 100 pounds and 300 watts due to the large size of the booster. This would allow computers such as the IBM 4Pi series, Autonetics D26J, and IBM's Saturn Launch Vehicle Digital Computer (LVDC) to be considered. Planners expected that the functions that later showed up on advanced Mariners-such as accelerometers, programmable sequencers with 512 words of memory, and telemetry registers-would be part of the proposed computer's capabilities and responsibilities. However, GE found that economies gained by a central system were outweighed by reliability advantages intrinsic to a distributed system¹.

Another approach came from Edward Greenberg, a Jet Propulsion Laboratory (JPL) engineer who programmed for the Mariner VI and VII Central Computer and Sequencer and contributed to the Viking Command Computer Subsystem (CCS) design. In December, 1972, he proposed that the Viking computer be standardized as a multimission processor². His intent was to reuse hardware and software development tools such as assemblers and simulators. Since one Viking computer could never handle all the functions needed on Voyager, several computers, each with a limited domain of functions, were needed.

Aside from the GE study and Greenberg's proposal, JPL [173] developed an additional argument for distributed computing. Edward H. Kopf, Jr., a JPL engineer specializing in attitude control, pointed out that different sections of the Laboratory needed computers to perform their assignments on Voyager and Galileo. Each group wanted its "own" computer, so that it would not be constantly competing for resources with other groups³. Therefore, a distributed system would help keep the peace.

The attractions of distributing computing, reliability, potential reusability, and separation of tasks, proved true in the development of the Voyager and Galileo spacecraft. Each has a functionally distributed set of computers. Voyager makes use of two of the Viking machines and a third, custom-built, computer. Each concentrates on processing different functions, such as attitude control, data formatting, and commanding. Galileo has dual processors for attitude control and six in a network for command and data handling. Both spacecraft were designed for long-duration, autonomous flight, a goal difficult to attain without the use of distribution.


Figure 6-1. The Voyager spacecraft with the radioisotope generators on the left boom and the scan platform on the right boom. (JPL photo P10727B)

^* Originally set for launch in the early 1980s, the mission slipped to May of 1986, but the grounding of the Shuttle fleet and cancellation of the Shuttle Centaur upper stage program in early 1986 led to an indefinite postponement and probably a change of launch vehicle.