V&V-based remaining fault estimation model for safety–critical software of a nuclear power plant

Quantitative software reliability measurement approaches have some limitations in demonstrating the proper level of reliability in cases of safety–critical software. One of the more promising alternatives is the use of software development quality information. Particularly in the nuclear industry, regulatory bodies in most countries use both probabilistic and deterministic measures for ensuring the reliability of safety-grade digital computers in NPPs. The point of deterministic criteria is to assess the whole development process and its related activities during the software development life cycle for the acceptance of safety–critical software. In addition software Verification and Validation (V&V) play an important role in this process. In this light, we propose a V&V-based fault estimation method using Bayesian Nets to estimate the remaining faults for safety–critical software after the software development life cycle is completed. By modeling the fault insertion and elimination processes during the whole development phases, the proposed method systematically estimates the expected number of remaining faults.

► A software fault estimation model based on Bayesian Nets and V&V.
► Use of quantified data derived from qualitative V&V results.
► Faults insertion and elimination process was modeled in the context of probability.
► Systematically estimates the expected number of remaining faults.