Fault-tolerant control of real-time systems in the presence of single event upsets

Abstract

Abstract High energy radiation environments, such as in space or in the presence of a nuclear event can adversely affect the operation of digital devices, presenting significant problems of reliable control to both Hardware and Software Engineers. These problems, although only recently addressed, have important implications for both military aircraft and space vehicles. The phenomenon known as Single-Event-Upset (SEU) occurs when bistable devices, such as memory, are corrupted by a collision with an high energy particle. These particles are found in abundance in radioactive environments and in outer space. An SEU can produce a variety of undesirable scenarios for a computer that range from annoying to catastrophic and can be either temporary or permanent. Some of these scenarios are; alteration of program memory, corruption of RAM or of a CPU register, and spurious or missed interrupts. A variety of techniques employing hardware, or both have been developed to enable recovery from an SEU. These schemes include RAM scrubbing, coding, fault-tolerant software, and judicious selection of hardware. A unique set of these and other techniques were developed for use on the Space Shuttle Inertial Measurement (IMU) computer. This paper provides an introduction to the phenomenon of the SEU, a review of some of the existing software techniques and hardware selection considerations to combat the effects of SEUs, and a discussion of the application of the techniques in the fault-tolerant, real-time control software for the Space Shuttle IMU's embedded computer.