A Holistic Approach to Functional Safety for Networked Cyber-Physical Systems

Functional safety is a significant concern in today's networked cyber-physical systems such as connected machines, autonomous vehicles, and intelligent environments. Simulation is a well-known methodology for the assessment of functional safety. Simulation models of networked cyber-physical systems are very heterogeneous relying on digital hardware, analog hardware, and network domains. Current functional safety assessment is mainly focused on digital hardware failures while minor attention is devoted to analog hardware and not at all to the interconnecting network. In this work we believe that in networked cyber-physical systems, the dependability must be verified not only for the nodes in isolation but also by taking into account their interaction through the communication channel. For this reason, this work proposes a holistic methodology for simulation-based safety assessment in which safety mechanisms are tested in a simulation environment reproducing the high-level behavior of digital hardware, analog hardware, and network communication. The methodology relies on three main automatic processes: 1) abstraction of analog models to transform them into system-level descriptions, 2) synthesis of network infrastructures to combine multiple cyber-physical systems, and 3) multi-domain fault injection in digital, analog, and network. Ultimately, the flow produces a homogeneous optimized description written in C++ for fast and reliable simulation which can have many applications. The focus of this thesis is performing extensive fault simulation and evaluating different functional safety metrics, \eg, fault and diagnostic coverage of all the safety mechanisms.