Certification of embedded systems: Quantitative analysis and irrefutable evidences

Abstract

Electronic/embedded systems are more and more dependent and relying on software flexibility and properties. They can be found in all spheres of our lives at a macro and global scale, ranging from personal and entertainment devices, household appliances, all types of transportation systems, global communication systems, civilian and military systems, energy and banking systems, and so on. Given the importance of all these systems and the safety and security requirements that become associated, national and international regulators require appropriate certification of each characteristic of the referred ubiquitous systems. This abstract presents the initial ideas concerning a quantitative analysis and evaluation of the evidence set forward in safety cases that support and eventually lead to certification of embedded systems with large parts of software. A discussion about the current industrial practices, limitations and state of the art related to certification evidences is drafted, and ideas concerning how can evidences be improved in terms of completeness, coherency, correctness, coverage, etc, as well as how can a quantitative analysis of the certification process be derived, are introduced for discussion and feedback. Current practices are not perfect, not properly applied, or applied in very different ways, presenting limitations, flaws and simplifications that put jeopardize systems safety, this is why we intend to initiate this research work.