A Combinatorial Approach for Exposing Off-Nominal Behaviors

Off-nominal behaviors (ONBs) have been a major concern in the areas of embedded systems and safety-critical systems. To address ONB problems, some researchers have proposed model-based approaches that can expose ONBs by analyzing natural language requirements documents. While these approaches produced promising results, they require a lot of human effort and time. In this paper, to reduce human effort and time, we propose a combinatorial-based approach, Combinatorial Causal Component Model (Combi-CCM), which uses structured requirements patterns and combinations generated using the IPOG algorithm. We conducted an empirical study using several requirements documents to evaluate our approach, and our results indicate that the proposed approach can reduce human effort and time while maintaining the same ONB exposure ability obtained by the control techniques.

1 Introduction

Off-nominal behaviors (ONBs) are unexpected or unintended behaviors of a system [1], [7], [12]. ONBs can occur for various reasons such as human errors caused by not following the intended procedures, components of a system being in conflicting states of operation, and environmental conditions resulting in such conflicting states. ONBs have been a major concern in the areas of embedded systems [1], medical devices, autonomous robotic systems [43], and safety-critical systems [2] because not addressing ONBs in these application domains might result in hazardous accidents and even catastrohpic results [29]. Unlike expected or nominal behaviors, ONBs requires knowledge acquisition of missing requirements and they are usually not specified in natural language (NL) requirements [12].