Skip to Main content Skip to Navigation

Formalisation et analyse algébrique et combinatoire de scénarios d'attaques généralisées

Abstract : The current definitions of a critical infrastructure are not adapted to the actual attacks which are observed these days. The problem is the same for the definition of an attack and therefore, the term « cyber attack » tends to reduce the conceptual and operational field of the person in charge of the security. Most of the approaches are reduced to identify the technical and IT domain only, and they forget the others domains specific to the intelligence. Then, the main methodologies to identify and to manage risk (EBIOS or some similar methodologies) take into account a definition of a critical infrastructure which is restrictive, static and local. The model of attacker and attack is also extremely narrowed as the technical approaches and the angles of attack of an attacker tend to be restricted to the IT domain only, even if the « cyber » angles may not exist or may only be a small part of an attack scenario.Therefore, it is necessary to have a new definition of a critical infrastructure, more complete and made according to the attacker point of view. Indeed, critical infrastructures can be protected by assessing the threats and vulnerability. This thesis aims to develop new models of infrastructure and attack accurately, models which will based on graph theory, with or without the cyber part. This graph-based representation is already used a lot to describe infrastructure, it will be enriched in order to have a more exhaustive view of an infrastructure environment. The dependencies with other entities (people, others critical infrastructures, etc.) have to be taken into account in order to obtain pertinent attack scenarios. This enriched representation must lead to new models of attackers, more realistic and implementing external components of the infrastructure which belong to its immediate environment. The main objective is the research of optimal paths or other mathematical structures which can be translated into attack scenarios. This global approach provides a finer (and therefore more realistic) definition of security as the lowest cost of the attack path.The research program is structured in five stages. The first two steps are aimed at defining the models and objects representing the security infrastructures as well as the attackers they are confronted with. The major difficulty encountered in developing a relevant infrastructure model is its ability to describe. Indeed, the more the model is rich, the more it can describe the infrastructure and the adversaries that attack it. The counterpart of developing a relevant model is its exponential characteristic. In these security models, we therefore expect that the problem of finding the vulnerabilities of a security infrastructure is equivalent to difficult problems, i.e. NP-hard or even NP-complete. The locks to be lifted will therefore consist in the design of heuristics to answer these problems in finite time with an ``acceptable" response. The third step is to define a generic methodology for assessing the safety of a security infrastructure. In order to validate the proposed models and methodology, the thesis program provides for the development of a research demonstrator in the form of an evaluation platform. Finally, the last step will be to evaluate an existing system from the platform by implementing the proposed methodology. The objective of this last step is to validate the models and the methodology and to propose an improvement if necessary.
Complete list of metadata

Cited literature [162 references]  Display  Hide  Download
Contributor : ABES STAR :  Contact
Submitted on : Monday, June 11, 2018 - 10:54:22 AM
Last modification on : Friday, October 23, 2020 - 5:04:44 PM
Long-term archiving on: : Wednesday, September 12, 2018 - 1:43:37 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01812052, version 1



Cécilia Gallais. Formalisation et analyse algébrique et combinatoire de scénarios d'attaques généralisées. Analyse numérique [math.NA]. Ecole nationale supérieure d'arts et métiers - ENSAM, 2017. Français. ⟨NNT : 2017ENAM0064⟩. ⟨tel-01812052⟩



Record views


Files downloads