Less Manual Work for Safety Engineers: Towards an Automated Safety Reasoning with Safety Patterns
This work addresses the problem of reducing manual work for safety engineers in deploying safety patterns, though it appears incremental as it builds on existing methods with limited automation.
The paper tackles the manual effort in safety-critical system development by proposing a logic programming approach to automate safety reasoning with safety patterns, demonstrating its application on automotive examples like adaptive cruise control and battery management systems.
The development of safety-critical systems requires the control of hazards that can potentially cause harm. To this end, safety engineers rely during the development phase on architectural solutions, called safety patterns, such as safety monitors, voters, and watchdogs. The goal of these patterns is to control (identified) faults that can trigger hazards. Safety patterns can control such faults by e.g., increasing the redundancy of the system. Currently, the reasoning of which pattern to use at which part of the target system to control which hazard is documented mostly in textual form or by means of models, such as GSN-models, with limited support for automation. This paper proposes the use of logic programming engines for the automated reasoning about system safety. We propose a domain-specific language for embedded system safety and specify as disjunctive logic programs reasoning principles used by safety engineers to deploy safety patterns, e.g., when to use safety monitors, or watchdogs. Our machinery enables two types of automated safety reasoning: (1) identification of which hazards can be controlled and which ones cannot be controlled by the existing safety patterns; and (2) automated recommendation of which patterns could be used at which place of the system to control potential hazards. Finally, we apply our machinery to two examples taken from the automotive domain: an adaptive cruise control system and a battery management system.