Multi-Scale Software Network Model for Software Safety of the Intended Functionality

Abstract

Software systems intensively interact with other software systems and hardware systems, and the potential hazards caused by the interaction with inadequate consideration becomes uncertain, especially the wide application of machine learning technology. Once the functions of software systems cannot meet the requirements of the interactions among software and hardware entities, safety problems caused by non-software system failures as software Safety of the intended functionality (SOTIF) arrise. The uncertainties of interation bring great challenges to SOTIF. In this paper, a multi-scale software network model is proposed based on complex network theory, and with the constructed network, test cases for software SOTIF can be efficiently generated. The key contribution is the uncertainties of interation among the software and hardware entities is digitally modeled, and can play a constructive role for guaranteeing SOTIF of software systems.