Distribute Real-time Fault-tolerance Studies On Coordination Control Information Processing Subsystem

Distribute Spacecraft System is made up of several micro satellites, and it can complete a complex task through coordination with each other. Coordination Control Information Processing Subsystem is in Distribute Spacecraft System, which must fit in with stance control, station control, communication managing and data disposing of DSS. There are some Distribute Real-time Fault-tolerance studies on CCIPS in this paper. At first, some relevant studies of Real-time Fault-tolerance on common system are present in this paper. Generally there are three kinds of resource redundancy in Fault-tolerance that are hardware redundancy, software redundancy, time redundancy and information redundancy. And there are two researche's fields about Real-time Fault-tolerance: one is Real-time Fault-tolerance Scheduling which comes from scheduling research; the other is Software Fault-tolerance Lib which comes from architecture research of Real-time Fault-tolerance system. Based on the analysis of Fault-tolerance and Real-time Fault-tolerance, this paper presents design and implementation of Real-time Fault-tolerance on CCIPS. When designing, we import "self-cure"and "immune"concepts. Every node of CCIPS can be regarded as an independent agent, and it has the capability of "self-cure"and "immune"to its own faults. Besides we synthesize node fault-tolerance and system fault-tolerance when designing Real-time Fault-tolerance on CCIPS. In node Fault-Tolerance, we use software method, hardware method, software and hardware method to take fault-tolerance that is about task of node and communication interface. And we make use of distribute system's traits to implement node-level fault-tolerance. Every node as an independent agent can collect system's information and make out judgment about system's state, and take fault-tolerance measure according to affording balance. At last, we do some tests of Real-time Fault-tolerance on CCIPS, which proves that Real-time task can complete under the conditions of task abnormity, communication's interface fault and node failure.