Design And Implementation Of Radiation-hardened Fault-tolerant Computer

Computer systems are often impacted by space radiation in aeronauticalarea and Aerospace. Especially the electromagnetic pulse poses a great threat tothe reliability of computer systems. It is difficult for human beings to repair thecomputer system because of the harsh environment, but the computer systemmust keep high availability. So it is important practical significant and economicvaluable to study how to improve the radiation-hardened ability of a computersystem. Traditional radiation-hardened technology can be used on chassis andchips. However electromagnetic pulse cannot be completely resisted. Theelectromagnetic pulse which couples to the chassis can also lead to hardwarefailure. So it is important to study how to improve the radiation resistance abilityof the entire system on the level of computer architecture.First of all, radiation effects and radiation-hardened technology were deeplystudied in this paper. According to fault-tolerant, the design of aradiation-hardened fault-tolerant computing system was proposed, which was amachine-level cold backup standby reserve system with only one backup. Theradiation resistance of a cold backup is better than hot ones, so cold backup andindependence were taken into account when designing detector and power supplycircuit.Second, CPLD logic and hardware circuit of the prototype were designedand implemented in detail. The prototype is consisted of two motherboards andone public board. Each motherboard has a fault management module. Thefault-tolerant management module detects faults in processor and I/O interfacecircuit where failures by radiation occur frequently. Resetting occurs when atransient fault is detected, and switching occurs when a permanent fault isdetected.In this paper，to verify the fault-tolerant mechanism of proposed system, animplantable fault injection system was developed, which could simulate theinjection of permanent and transient faults that happened in the processor or I/Ointerface. At last faults were injected into the prototype by the implanta ble fault injection system. The experimental result shows that the fault detectionmechanism and fault handling capacity of the computer system can meet demand.