Séminaire confiance numérique (Azzam Mourad, LAU)

Security issues and Directions of Intelligent Transport Systems within limited-resources constraints

Intelligent transport system (ITS) is becoming a necessity in our everyday life for traffic control, time management, real-time traffic flow, incident management, etc. In the context of limited-resources constraints (i.e. lack of infrastructure), some necessary information may not be available and/or a single source of data may not be enough in order to provide accurate decision by the core modules. In this context, the proposed idea is to rely on vehicles as principal source of information, diversify the sources of data and build cooperative and service-based models at all layers of ITS in order to target the aforementioned problem. However, such approach entails several challenges in terms of adaptability, context-awareness, QoS and security. In my talk, I will be discussing security problems and solutions targeting separately services cooperation/composition and vehicular networks, which will potentially form the base to elaborate the intended approach.

Starting with the security of services cooperation, current solutions for Web services composition (e.g. BPEL) still do not provide mechanisms to dynamically adapt the business process with the rapid changes in business goals and requirements. Therefore, more efforts are needed to reach high adaptation of composition at several levels, most importantly, business logic, behavioral rules and security policies. In terms of security, XACML (eXtensible Access Control Markup Language) has been known as the de facto standard widely used by many vendors for specifying access control and context-aware policies for individual Web services. Due to the complexity of interaction in composite services, diversity of rules, and notably complexity of policies specification languages, many flaws and conflicts are more likely to arise, specially that the size of XACML policies are significantly growing to cope with the composed behaviors. In this context, we will discuss E-AspectBPEL, a model driven approach based on aspect-oriented programming for defining both context-aware policies to reach higher adaptability, and dependability between policies hardened in the BPEL process to overcome the conflict among them. Moreover, we will discuss SBA-XACML, a novel set-based algebra scheme that provides better evaluation performance and allows to detect flaws, conflicts and redundancies in XACML policies.

As for the security of Vehicular network, we will discuss the problem of clustering in Vehicular Ad Hoc Networks (VANETs) in the presence of selfish nodes. The problem arises due to the highly mobile topology of VANET, which leads to frequent clusters disconnections and alters the Quality of Service (QoS) over the network. Moreover, the presence of selfish nodes would hinder the application of any clustering model. These nodes behave rationally and tend hence to maximize their gain regardless of the negatives implication that may affect the network. Thus, the proposed solution is concerned with introducing a stable clustering model that deals with the high mobility of vehicles by considering a tradeoff between QoS requirements and high mobility constraints. This model is accompanied with a multi-phase model able to stimulate the cooperation of the vehicles, detect the selfish nodes, and regulate the cooperation. In this regard, several theories and algorithms will be discussed such as Ant Colony Optimization algorithms, incentive mechanism, cooperative watchdog based on Dempster-Shafer theory and repeated game theory based on Tit-for-Tat strategy.

537 vues

Vendredi 13 Juin 2014

Thème(s) : Conférences Recherche

Tag(s) : confiance numerique, sécurité informatique

Intégrer :

Vidéos suggérées

Erreur !