The SystemX Technological Research Institute is located in the heart of the scientific campus of Paris-Saclay, and it develops world-class technological research in the field of digital systems engineering. In close interaction with the best French research organizations in the field and thanks to mixed teams of scientists and engineers with industrial and academic background, IRT SystemX’s mission is to generate new knowledge and technological solutions based on breakthroughs in digital engineering and to disseminate its skills to all economic sectors.
Within IRT SystemX, the PhD student will be attached to the « Data Science and Interaction » scientific axis. The thesis topic has been defined by the consortium brought together in the framework of the « Trusted Autonomous Mobility » (TAM) project. The TAM project aims at defining new cybersecurity solutions for cooperative transport systems.
The thesis will be supervised by Fawzi Nashashibi within the RITS team at INRIA (Paris) and will be registered at the doctoral school ISMME n°621 (Ingénierie des Systèmes, Matériaux, Mécanique, Energétique). The doctoral student will also benefit from scientific supervision at the IRT SystemX by Ines Ben Jemaa, research engineer within the TAM project.
The position is based at the IRT SystemX – Palaiseau. The PhD student will visit frequently INRIA (Paris) and will participate to internal activities of the RITS team.
If the pandemic context continues, the PhD student will follow recommendations of the IRT SystemX management (for example, in case the recommendation is to work 100% remotely).
IRT SystemX will provide the necessary hardware and software and in general the necessary equipment for the PhD student (laptop, VPN access, etc).
Cooperative Intelligent Transport Systems (C-ITS) rely on wireless communications to ensure road safety. Several ITS services enable road safety use cases by the exchange of standardized messages among the ITS communicating entities, which may be vehicles or infrastructure nodes called Road Side Units (RSU). Collective perception is a new ITS service, based on the exchange of lists of perceived objects among the ITS communicating entities. The list of perceived objects in the road environment is established by vehicles and RSUs using data gathered by embedded sensors.
Intersection crossing is a typical example that illustrates the usefulness of this service. Vehicles approaching the intersection as well as RSUs deployed in the intersection periodically broadcast Collective Perception Messages (CPM). Each CPM contains the list of objects perceived by the emitting ITS station thanks to its embedded sensors. When receiving a CPM, a vehicle or a RSU is able to extend its awareness of the environment by updating its own perception list with the objects contained in the neighbors’ (other vehicles or RSUs) CPMs. As a result, a vehicle is able to anticipate potential dangerous situations at the intersection, and thus cross the intersection in a secure and safe way.
CPMs may however contain abnormal perception data, revealing a potential misbehaving sender. Misbehavior is classified into two types. The first one is intentional misbehavior and occurs when a malicious vehicle or RSU broadcasts false information to disrupt and attack the ITS system. The second type of misbehavior is non-intentional and occurs when an embedded sensor or its associated perception system is faulty and, consequently, provides erroneous data. Detecting misbehavior of the communicating entities is a crucial functionality of cybersecurity for the C-ITS, and specifically allows setting up the appropriate counter-measures to avoid risky situations.
Objectives and scientific challenges
The work will be divided in two parts. The first concerns local detection of misbehavior by vehicular network nodes. For each vehicle and RSU this operation consists on a preliminary auto-diagnostic step, where each entity seeks inconsistencies or non-plausibility in their own sensor data to detect potential malfunctioning devices. After the auto-diagnostic step, vehicles and RSU process perception data received from other nodes, checking if there is any inconsistency revealing a security attack or a faulty device. The objective of this part of the work is to propose reliable and robust local misbehavior detection solutions, which take into consideration vehicular communications requirements and constraints. It would be relevant to explore the use of existing approaches such as probabilistic approaches (e.g, Bayesian inference, etc), evidence theories (e.g, belief theory, etc) and knowledge approaches (e.g, machine learning, fuzzy logic, etc).
The second part of the work focusses on global misbehavior detection. Indeed, a misbehavior authority, localized in the ITS backend security system, receives cybersecurity alerts whenever a vehicle or RSU locally detects a misbehavior in the network. The misbehavior authority is assumed to have great data storage and data processing capacities to handle the huge amount of local misbehavior alerts. The objective of the global detection is to enhance misbehavior detection quality compared to local detection (for instance by reducing the false positive rate). The objective of this part is to investigate the best data analysis approaches including machine learning approaches suitable for misbehavior detection on the collective perception service, to provide reliable and robust global detection results.
A previous project (SCA for “Secure Cooperative Autonomous Systems”) implemented F2MD (Framework For Misbehavior Detection), an extension of the Veins simulator for misbehavior detection in C-ITS. The PhD candidate will implement, validate and evaluate the proposed solutions within this simulation platform. If time allows, the proposed solutions could be totally or partially integrated in a Proof Of Concept prototype in the TAM project for real experimentation purposes.
- W. van der Heijden, S. Dietzel, T. Leinmüller and F. Kargl, « Survey on Misbehavior Detection in Cooperative Intelligent Transportation Systems, » in IEEE Communications Surveys & Tutorials, vol. 21, no. 1, pp. 779-811, Firstquarter 2019, doi: 10.1109/COMST.2018.2873088.
- W. van der Heijden, F. Kargl “Evaluating Misbehavior Detection for Vehicular Networks” 5th GI/ITG KuVS Fachgespräch Inter-Vehicle Communication (FG-IVC 2017), Erlangen
- ETSI TS 103 324 V0.0.20 (2021-02) Specification of the Collective Perception Service
- Kamel, A. Kaiser, I. Ben Jemaa, P. Cincilla, and P. Urien, “CaTch: a confidence range tolerant misbehavior detection approach”. IEEE Wireless Communications and Networking Conference (WCNC), Marrakech, Morocco, 2019
- Allig, T. Leinmüller, P. Mittal and G. Wanielik, « Trustworthiness Estimation of Entities within Collective Perception, IEEE Vehicular Networking Conference (VNC), Los Angeles, USA, 2019.
- K. Garlichs, H. Günther and L. C. Wolf, « Generation Rules for the Collective Perception Service, » 2019 IEEE Vehicular Networking Conference (VNC), Los Angeles, CA, USA, 2019, pp. 1-8, doi: 10.1109/VNC48660.2019.9062827
- J. Kamel, M. R. Ansari, J. Petit, A. Kaiser, I. B. Jemaa and P. Urien, « Simulation Framework for Misbehavior Detection in Vehicular Networks, » in IEEE Transactions on Vehicular Technology, vol. 69, no. 6, pp. 6631-6643, June 2020, doi: 10.1109/TVT.2020.2984878.
Candidate must hold a master or engineering degree in information processing, computer science, telecommunication
Technical and professional skills :
- Statistical analysis, probabilities, machine learning
- C++, Python, Linux OS, Latex
- Fluent English
- Knowledge in C-ITS cybersecurity is appreciated
- Dynamic, proactive, well organized, strong team working skills
REF : DIT-2021-01
Type of contract : Thesis
Location of the post : Cluster Paris Saclay (91)
Desired starting date : September, 1st 2021
Application documents : Detailed CV, cover letter, scores of the last year, reference contacts, recommendations letters are appreciated