A Natural Language Processing (NLP) Framework for Embedded Systems to Automatically Extract Verification Aspects from Textual Design Requirements

Abstract

Embedded systems requirements are significantly different with respect to general purpose systems due to the safety-critical nature and the presence of temporal aspects. Particularly, the design requirements of embedded systems, comprise several temporal conditions, are first identified. Subsequently, a test engineer / system engineer analyzes the design requirements manually to identify the verification characteristics and develops the verification assertions / constraints accordingly. However, the manual analysis of design requirements for verification is time consuming task. Furthermore, high level of domain expertise is required to develop the correct and complete verification assertions from the design requirements. This article presents a novel Natural Language Processing (NLP) framework for embedded systems to analyze and automatically extract verification aspects from the textual design requirements. This leads to considerably simplify and accelerate the development of verification assertions. As a part of research, a complete AR2AA (Automated Requirements 2 Assertions Analyzer) tool is developed in C# by utilizing the SharpNLP and regular expression libraries. The usefulness of proposed framework is demonstrated through Car Collision and Avoidance System (CCAS) case study. The initial results prove that the proposed framework is highly effective for the analysis and development of verification assertions from the textual design requirements.