Séminaire 2014

Séminaire "Modeling and Analysis of Critical Systems"

13 juin 14h30-16h30 - salle BL005 à TELECOM SudParis à Evry

Title : Modeling and Analysis of Critical Systems

Speaker : Nejib Ben Hadj-Alouane, Professor in Computer Engineering, National School of Engineers of Tunis (ENIT), The OASIS Laboratory (Optimization, Analysis of Industrial Systems and Services)

Abstract :

Our work on critical systems targets a number of problem areas and applications, all requiring the development of models for analysis and verification, and tools for decision making. This talk focuses on three different problem areas, each requiring a different modeling framework and solution approach. Throughout our talk we discuss several research perspectives and applications linked to modern technologies, such as Web Services and Cloud computing.

We begin by exposing our new Petri nets modeling framework, Extended Timed Petri Nets (ETPN), useful in modeling a special class of hybrid systems, with a weak continuous component and a strong discrete one. These types of hybrid systems are commonly encountered when dealing with modern man-made, embedded real-time systems. ETPNs allow for capturing global, constant speed, variables (representing time and/or continuous quantities), and facilitates dealing with interrupting and restarting tasks. We discuss a supervisory control problem based on this model and other related issues, dealing with model transformation and complexity.

Our second problems deals with operating high-demand and high-performance virtualized data centers (DCs). We focus on the development of tools, based on operations research models and techniques, for the management of theses DCs, while striving to improve user applications performance and productivity. In fact, performance metrics such as propagation delay and Quality-of-Service (QoS) are important metrics for many end-user applications and services (e.g., video streaming servers, data intensive applications, etc.).

Our work on virtualized DCs, focuses on the problem of virtual machines (VMs) placement in geographically distributed data centers. We consider communicating VMs assigned to data centers that are connected via a backbone network. We aim to plan and optimize the placement of VMs in data centers so as to minimize the IP-traffic within the backbone network.
Thus, we propose first, a formulation which can be considered as a variant of the Hub Location problem modeling and we show its extreme difficulty for medium and large size instances. In order to overcome this difficulty, we reformulate the problem by multi-commodity flow, adopt variable aggregating methods and add valid inequalities to strengthen this new formulation. The different experiments that we present show the effectiveness of our last model in terms of running time and computational resources.

Our second problem deals with security and in computer systems, services and protocols. We introduce the property of opacity and show how it is used to capture secrecy-related problems. We focus on the development of an on-line method for the efficient verification of opacity in models based on automata.