Security for Complex Systems

Aller à : navigation, rechercher


Information systems are built of myriads of independent, interconnected devices, organized in loosely interdependent sub-systems (enterprise networks or site, public access networks, etc.). These information systems contain organization-critical data such as financial or commercial statements, personal devices, as well as critical industry systems known as SCADA (electrical grid, control systems for nuclear plants or aviation control, etc.). They are interconnected with various levels of control, and various constrains on confidentiality, integrity, availability. We make the assertion that this complexity is a domain of choice for defining security models based on an abstracted view of the systems, taking into account their heterogeneity and considering, when useful, low level implementation details. In particular, emerging risks are best such as interconnection between partner organization, or access to critical industry systems, are best considered on a systemic level that through low level, non-scalable models. Research in the domain of security for complex system will be conducted using two complementary target systems:

  • Information systems as complex systems: controlling access to resources, propagation of access rights, and cohabitation of heterogeneous systems: core servers, domain computers, external computers (e.g. over Wifi), mobile devices (BYOD)
  • Social networks as complex systems: data protection, data propagation and data analysis

The life-cycle phases of the security analysis for complex systems are:

  • Risk-driven analysis: which are the main dangers in the system?
  • Security at the architectural level: do security patterns help, and are they used in a consistent manner
  • Security at the implementation level: do the code and configuration comply with intended security properties?
  • Log mining for post-mortem studies: how do the execution/data exchange trace enable to characterized security leaks or attacks?

Use cases

The issue of security for complex system is investigated in the scope of following projects:

  • ANR EASEA-Cloud: EASEA-Cloud is a platform for performing distributed analysis of the behavior of complex systems through evolutionary algorithms. Security is here seen as a technical enabler for protecting valuable data from industry partners. The EASEA-Platform can also be seen as a simplified view of an information system, which is itself a complex system composed of atomic entities exhibiting rich interactions. EASEA provides therefore a prototypical view of the Information System as a complex system
  • Complex system MOOC using Man-Hill: Man-Hill act as a specialized social network where teachers and students provide training and exercise data, and where the progression is autonomously adapted according to the 4P approach; Participatory (each actor of the system is able to provide input or help), Personalized (curriculum is evolving according to student results), Predictive (according to student curriculum, gaps and requirements of specific training can be identified), Preventive (Pitfalls can be documented and focused on). Security issues consist in guaranteeing the propagation of data and in enforcing actor roles.
  • 4P medicine: 4P medicine requires building contact between patients as well as between patients and practitioners. As far as it implies sharing personal as well as medical data, it brings with it severe ethical and technological challenges: which data can be shared with professionals, with other patients? Which data can be made available for analysis, under which conditions? One tool identified to answer the functional requirements of 4P medicine is social networking. We will therefore focus our analysis on the security issues specific to social networks and their implementation in this specific context.
  • SME Information system: besides the EASEA-Cloud Infrastructure, the Information System of the ECAM Strasbourg-Europe engineering school provides a second simple example of an enterprise network. Security politics and their implementation will be used for validating the security model defined in the context of the EASEA-Cloud project.


The ECAM Strasbourg-Europe has a project of deploying EASEA-Cloud on a computing room used for teaching, so as to complement existing resources at the University of Strasbourg and to leverage existing resources.