An Introspection Framework for Fault Tolerance in Support of Autonomous Space Systems

This paper describes a software system designed for the support of future autonomous space missions by providing an infrastructure for runtime monitoring, analysis, and feedback. The objective of this research is to make mission software executing on parallel on-board architectures fault tolerant through an introspection mechanism that provides automatic recovery minimizing the loss of function and data. Such architectures are essential for future JPL missions because of their increased need for autonomy along with enhanced on-board computational capabilities while in deep space or time-critical situations. The standard framework for introspection described in this paper integrates well with existing flight software architectures and can serve as an enabling technology for the support of such systems. Furthermore, it separates the introspection capability from applications and the underlying system, providing a generic framework that can be also applied to a broad range of problems beyond fault tolerance, such as behavior analysis, intrusion detection, performance tuning, and power management.