Font Size: a A A

Real-time Fault-tolerant Embedded Small Satellite Operating System Design

Posted on:2001-03-01Degree:DoctorType:Dissertation
Country:ChinaCandidate:T LiFull Text:PDF
GTID:1112360002450794Subject:Microelectronics and Solid State Electronics
Abstract/Summary:PDF Full Text Request
Relevant studies on the design of micro-satellite embedded real-time faulttolerant and regroupable operating system and further research in relevant fields of software engineer are presented in this dissertation.The structural design of micro-satellite operating system in the space condition is discussed and researched in detail at first. After the comparison of lots of operating system models, micro-satellite operating system design philosophy based on micro kernel is introduced. This model decrease the inter-module calls sharply, thereby, assure the high reliability and extensibility from the structure.Subsequently, the scheduling strategy of the system processors is discussed. After studying time driven and event driven and considering the specific demands of microsatellite operating system, time-event driven preemptive scheduling strategy based on message processing is chosen to meet the system demands. This strategy synthesis the benefits of both event driven and time driven, could lower the response time of critical task and avoid losing control when a suddenly overload appears. Furthermore, the partition of normal message and periodical task message, decr鏰se sharply not only the CPU cost of the scheduler but also the complexity of inner program and improve the flexibility and extensibility of the system. Mechanism of injection management, the introduction of it give the system the ability to rer~ew a process to implement process regroup when fault is discovered or introduce new enhanced function to the system, open another door to improve the reliability of system.Fault-tolerance and recovery is a classic emphasis of micro-satellite operating system design. Taking account of this, discuss of fault-tolerance and recovery of micro-satellite operating system is introduced and a new kind of method to identify fault-prone satellite software modules based on modified cascaded-correlation learning architecture is presented. Comparing with other possible methods such as sense machine method and Gauss method, this method can identify fault-prone modules of satellite software in early phase of software development according to complexity indexes of each module with fast convergence arid better-result expectability. This is very useful to assign the designing, developing and debugging workload in reason, to improve the reliability of micro-satellite software system pertinently and to decrease software development cost and the times of repeated rewriting of code in the later period apparently.Computer duplex is used in satellite computer system to improve the reliability. How to develop a method to synchronize timers and processors of these twocomputers and provide an efficient asynchronous communication interface and mutex and synchronization interface between different processes becomes one of the main task of this dissertation. A new kind method to provide distributed mutex and synchronization service based on mutex and synchronization service processes located on each station is presented. Not limited to satellite operating system, it can be referred in all kinds of distributed systems.Another problem that must be studied in computer duplex system or multicomputer system is the problem of task assignment. Commonly, tasks in microsatellite system run in predefined mode without any stochastic and outburst behavior which often happen in general computer systems on the ground. Considering of this, quasi-static task assignment algorithm is considered a good algorithm to allocate tasks and keep balancing of overload with powerful value. A new kind of static task assignment algorithm based on Hopfield Network is presented in this dissertation. According to the application result, the algorithm is fast convergent and better result expectable. Furthermore, fault process migration mechanism is studied and introduced to assure system could backward recovery rapidly to a consistent state when one of the duplex is crashed.
Keywords/Search Tags:Fault-tolerant
PDF Full Text Request
Related items