Using Runtime Quantitative Verification to Provide Assurance Evidence for Self-Adaptive Software - Advances, Applications and Research Challenges

Providing assurance that self-adaptive software meets its dependability, performance and other quality-of-service (QoS) requirements is a great challenge. Recent approaches to addressing it use formal methods at runtime, to drive the reconfiguration of self-adaptive software in provably correct ways. One approach that shows promise is runtime quantitative verification (RQV), which uses quantitative model checking to reverify the QoS properties of self-adaptive software after environmental, requirement and system changes. This reverification identifies QoS requirement violations and supports the dynamic reconfiguration of the software for recovery from such violations. More importantly, it provides irrefutable assurance evidence that adaptation decisions are correct. In this paper, we survey recent advances in the development of efficient RQV techniques, the application of these techniques within multiple domains and the remaining research challenges.