Tools for Software Safety Analysis in Aerospace Applications

This proposal concerns the development of procedures and software for implementing the Dynamic Flowgraph Methodology (DFM), a methodological approach to modeling and analyzing software-based control systems for the purpose of reliability/safety assessment and verification. Using DFM, models representing causal and timing relationships between software functions, interfacing hardware and external system parameters are analyzed to produce "timed" fault trees which relate the values of system variables at discrete points in time. The development of dynamic, multi-valued (non- binary) fault trees represents a significant technical advancement in system safety and reliability analysis. These fault trees explicitly identify the causal and timing effects leading to system events of interest. This information can then be used to eliminate system faults resulting from combinations of software logic errors, hardware failures and environmental conditions. The use of the tools developed by this research, instead of traditional testing and verification techniques, will enable NASA and its contractors to save large amounts of effort and economic resources when demonstrating the safety of digital systems and software.