Completeness in Software Requirements Specifications for Process Control Systems

这项工作的目标是为确定给定的一组软件需求是否在内部完成提供严格的基础。这将需要确定提供完整的功能和性能规范所需的个别需求的类型，以及允许为派生需求制定推理规则和结束标准的正式符号。无论是出于经济原因还是安全原因，这个问题都很重要。在软件开发过程的后期阶段纠正需求规范错误的高成本和高难度意味着，通过帮助及早发现问题的方法可以获得巨大的影响力。此外，大量与软件相关的事故可以追溯到需求问题，而需求方面的不完整性通常是一个偶然的因素。此外，安全软件行为的正式分析不能仅通过测试来保证；对系统的指定行为的分析是必要的，如果行为规范不完整，这是不可能的。重点将放在软件需求规范以前没有得到充分处理的方面，包括时序抽象、人机界面和健壮性。

The goal of this work is to provide a rigorous basis for ascertaining whether or not a given set of software requirements is internally complete. This will require determining the types of individual requirements that are necessary to provide complete functional and performance specifications along with a formal notation that will allow the development of rules of inference and closure criteria for derived requirements. This problem is important both for economic and safety reasons. The high cost and difficulty of correction requirements specification errors in the later stages of the software development process implies that great leverage can be obtained by methods that help to find problems early. Also, a large number of software-related accidents can be traced to requirements problems, and incompleteness in terms of requirements is often a casual factor. Furthermore, formal analysis of safe software behavior cannot be guaranteed by testing alone; analysis of the specified behavior of the system is necessary and this is not possible if the behavioral specification is incomplete. Emphasis will be placed on aspects of software requirements specification that have not previously been adequately handled including timing abstractions, the man/machine interface, and robustness.