Visible to the public Using Statistical Methods and Co-Simulation to Evaluate ADS-Equipped Vehicle Trustworthiness

TitleUsing Statistical Methods and Co-Simulation to Evaluate ADS-Equipped Vehicle Trustworthiness
Publication TypeConference Paper
Year of Publication2019
AuthorsHalba, Khalid, Griffor, Edward, Kamongi, Patrick, Roth, Thomas
Conference Name2019 Electric Vehicles International Conference (EV)
KeywordsAcceleration, ADS-equipped vehicle trustworthiness, ADS-equipped vehicles, application program interfaces, automated driving system-equipped vehicles, Co-Simulation, complex cyber-physical systems, composability, Computational modeling, Cyber-physical systems, driver information systems, FMI, fractional factorial design, functional mock-up interface, Internet, Measurement, middleware, middleware security, multiple simulation platforms, NIST, NIST framework, OpenModelica, policy-based governance, pubcrawl, resilience, Resiliency, Ricardo IGNITE, Safety, safety metrics, security of data, Sensors, simulated operational domain design, system reliability, Tools, trustworthiness, UCEF, vehicle dynamics simulators, virtual analysis functions, virtual reality
Abstract

With the increasing interest in studying Automated Driving System (ADS)-equipped vehicles through simulation, there is a growing need for comprehensive and agile middleware to provide novel Virtual Analysis (VA) functions of ADS-equipped vehicles towards enabling a reliable representation for pre-deployment test. The National Institute of Standards and Technology (NIST) Universal Cyber-physical systems Environment for Federation (UCEF) is such a VA environment. It provides Application Programming Interfaces (APIs) capable of ensuring synchronized interactions across multiple simulation platforms such as LabVIEW, OMNeT++, Ricardo IGNITE, and Internet of Things (IoT) platforms. UCEF can aid engineers and researchers in understanding the impact of different constraints associated with complex cyber-physical systems (CPS). In this work UCEF is used to produce a simulated Operational Domain Design (ODD) for ADS-equipped vehicles where control (drive cycle/speed pattern), sensing (obstacle detection, traffic signs and lights), and threats (unusual signals, hacked sources) are represented as UCEF federates to simulate a drive cycle and to feed it to vehicle dynamics simulators (e.g. OpenModelica or Ricardo IGNITE) through the Functional Mock-up Interface (FMI). In this way we can subject the vehicle to a wide range of scenarios, collect data on the resulting interactions, and analyze those interactions using metrics to understand trustworthiness impact. Trustworthiness is defined here as in the NIST Framework for Cyber-Physical Systems, and is comprised of system reliability, resiliency, safety, security, and privacy. The goal of this work is to provide an example of an experimental design strategy using Fractional Factorial Design for statistically assessing the most important safety metrics in ADS-equipped vehicles.

DOI10.1109/EV.2019.8892870
Citation Keyhalba_using_2019