Abstract
It does not make sense to grant carte blanche high-assurance certificates to product that may be used across multiple platforms and in multiple environments. We should bind software certification to a product's known environment and operational profile. The author proposes three techniques for verifying high assurance: desirable-behavior testing, abnormal testing, and fault injection. Each uses the product's operational profile to detect software-related anomalies that might allow a catastrophic event

This publication has 7 references indexed in Scilit: