Systems integration requires verificationVerification at each level of integration beginning at the atomic part and building its way to full system-level verificationVerification. Traditionally, the development of a verification strategyVerification strategy—which determines when and how to perform these activities—relies on heuristicsHeuristics and best practices established early in the system’s lifecycle. However, executing verificationVerification activities is a significant cost driver in systems integration. Both industry and academia are seeking methods to optimize verification strategiesVerification strategy by eliminating excessive activities without diminishing confidence in system performance. To achieve such optimization, we need metrics that effectively capture the inherent value of verificationVerification. While various metrics have been proposed, none have offered an approach intrinsic to the verificationVerification process itself. This paper proposes that concepts from information theoryInformation Theory—specifically entropyEntropy and information gain—may provide suitable metrics for quantifying the fundamental value of verificationVerification.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Understanding the Value of Verification

  • Jack Fitzpatrick,
  • Hanumanthrao Kannan

摘要

Systems integration requires verificationVerification at each level of integration beginning at the atomic part and building its way to full system-level verificationVerification. Traditionally, the development of a verification strategyVerification strategy—which determines when and how to perform these activities—relies on heuristicsHeuristics and best practices established early in the system’s lifecycle. However, executing verificationVerification activities is a significant cost driver in systems integration. Both industry and academia are seeking methods to optimize verification strategiesVerification strategy by eliminating excessive activities without diminishing confidence in system performance. To achieve such optimization, we need metrics that effectively capture the inherent value of verificationVerification. While various metrics have been proposed, none have offered an approach intrinsic to the verificationVerification process itself. This paper proposes that concepts from information theoryInformation Theory—specifically entropyEntropy and information gain—may provide suitable metrics for quantifying the fundamental value of verificationVerification.