IRM

Due to progressive depletion of IPv4 addresses, the new IPv6 protocol has been created, and gets progressively deployed nowadays. In the IPv4 world, it is usual for companies to get multihomed, that is, asking connectivity to several Internet Service Providers. This gives such companies better reliability and flexibility for their Internet traffic. But IPv6, while solving the problem of address shortage, comes also with its set of new challenges. In particular address allocation policies were intended to be radically changed, with the consequence for multihomed sites to have to manage several addresses per computer.
As an answer to this new landscape, the IETF designed the Shim6 protocol, aimed at providing an environment for multihoming management, while moving multihoming control from network to end-hosts.
The goal of this presentation is really tutorial. We will present the why and the how of Shim6, as well as its implications and drawbacks.

Les systèmes de coordonnées sont des systèmes distribués ayant pour but, à partir de mesures de distance (par exemple RTT) entre certaines paires de noeuds, d’associer des coordonnées à chaque noeud dans un espace métrique. L’idée principale est que si chaque noeud peut être associé à des coordonnées virtuelles dans un espace approprié, la distance entre les noeuds est trivialement calculée sans pour autant avoir recours à des mesures directes. L’avantage premier de ces systèmes est que si les distances réseaux (dans le sens de délais) sont plongées (''embedding'') dans un espace de coordonnées où une position raisonnablement précise pour chaque noeud est établie, le surcoût de mesures produit par le positionnement, est amorti sur plusieurs prédictions de distances. Ceci réduit énormément le coût en termes de mesures de distances du système entier. Dans ce tutoriel, on étudiera les différentes approches qui existent: les techniques basées sur les des hôtes références (GNP, Lighthouses, etc) et les techniques distribuées (Vivaldi, PIC, etc). On montrera aussi les limitations de ces techniques.

Comme chaque année, l'équipe sysadmin fait le point avec vous sur l'état de l'infrastructure informatique en INGI. Après une mise en jambes présentant l'état de nos différents services, nous exposerons nos réalisations de l'année écoulée, l'état d'avancement de nos projets en cours, ainsi que les évolutions futures envisagées pour notre système informatique. Nous espérons aussi que ce sera l'occasion d'échanger avec vous impressions, commentaires et désidératas à propos de notre outil de travail commun.

<Abstract of presentation.>

Les systèmes de coordonnées sont des systèmes distribués ayant pour but, à partir de mesures de distance (par exemple RTT) entre certaines paires de noeuds, d’associer des coordonnées à chaque noeud dans un espace métrique. L’idée principale est que si chaque noeud peut être associé à des coordonnées virtuelles dans un espace approprié, la distance entre les noeuds est trivialement calculée sans pour autant avoir recours à des mesures directes. L’avantage premier de ces systèmes est que si les distances réseaux (dans le sens de délais) sont plongées (''embedding'') dans un espace de coordonnées où une position raisonnablement précise pour chaque noeud est établie, le surcoût de mesures produit par le positionnement, est amorti sur plusieurs prédictions de distances. Ceci réduit énormément le coût en termes de mesures de distances du système entier. Dans ce tutoriel, on étudiera les différentes approches qui existent: les techniques basées sur les des hôtes références (GNP, Lighthouses, etc) et les techniques distribuées (Vivaldi, PIC, etc). On montrera aussi les limitations de ces techniques.

Due to progressive depletion of IPv4 addresses, the new IPv6 protocol has been created, and gets progressively deployed nowadays. In the IPv4 world, it is usual for companies to get multihomed, that is, asking connectivity to several Internet Service Providers. This gives such companies better reliability and flexibility for their Internet traffic. But IPv6, while solving the problem of address shortage, comes also with its set of new challenges. In particular address allocation policies were intended to be radically changed, with the consequence for multihomed sites to have to manage several addresses per computer.
As an answer to this new landscape, the IETF designed the Shim6 protocol, aimed at providing an environment for multihoming management, while moving multihoming control from network to end-hosts.
The goal of this presentation is really tutorial. We will present the why and the how of Shim6, as well as its implications and drawbacks.

Pseudo-random sequences have many applications in communication including radar ranging, scrambling or stream ciphers. Many of those contexts require a higher throughput or a lower power consumption than the one expected by the single symbol/cycle PRNG. Better performances are achieved by using a decimated representation of the given sequences. I will make an overview of the advances in this field in the case of m-sequences and l-sequences.

Introduction to Mozart-Oz and its concurrency model. It will be a 2 hours tutorial (a bit more maybe). You don't need to know anything about Mozart-Oz to attend. We will go through the basic declarative model, and how to extend it to a declarative concurrent model with data flow synchronization. Then we add message passing to move forward to basic distributing programming. If we have time, we can see some shared state concurrency.

L'identification par radio-fréquence est une technologie sans contact utilisée pour identifier et/ou authentifier des objets ou personnes isolés à travers un canal par ondes radio grâce à deux dispositifs : lecteur et tag RFID. Concrètement un tag sera mis à l'intérieur d'un objet et celui-ci sera interrogé quand il sera dans le champs d'action électromagnétique d'un lecteur : le tag devra mener à bien une authentification demandée par le lecteur, c'est-à-dire qu'il devra prouver son identité au lecteur.
De nombreuses applications utilisent maintenant la RFID, tels que la billetterie des transports publics, le contrôle d'accès, ou encore la gestion des stocks.
Et bien que la RFID possède un nombre considérable d'atouts, des problèmes majeurs sont à prendre en considération pour l'authentification RFID :
- La cryptographie à bas coût
- La privacy
- Les attaques par relais
- Les lecteurs compromis
Notre préoccupation principale ici concerne ce dernier point, car il est souvent laissé de coté dans la littérature scientifique. Par définition, on considère qu'un lecteur d'un système est compromis s'il tombe entre les mains d'un attaquant, c'est-à-dire qu'un attaquant possède toutes les informations stockées dans le lecteur corrompu.
Le but de ces travaux de recherche est de proposer des protocoles d'authentification RFID qui puissent faire face à ce problème des lecteurs compromis. En fait si un lecteur a été piraté, la sécurité n'existe plus. Donc, la principale propriété de nos solutions est de pouvoir rétablir la sécurité du système sans avoir besoin de remplacer physiquement les appareils (essentiellement les tags).

During the past ten years, we saw the emergence of a set of applications requiring more and more quality of service (QoS). While in the past content was located in a single place, it is, now, frequent that the content is replicated among a set of servers located anywhere in the Internet or among users. Examples of this are peer-to-peer (P2P) applications and FTP mirrors. In addition, thanks to the multihoming, i.e., the ability to have different connections to Internet through different providers, leveraging path diversity becomes important. The new complexity of the Internet motivate the Routing Research Group (RRG) of the Internet Research Task Force (IRTF) to discuss several architectural solutions to build an interdomain routing architecture that faces these new challenges.
During the seminar, we will discuss the main problems to be tackled by the future Internet design. We will then present the locator/identifier separation paradigm and see how it can help to the evolution of the Internet. Next, we will see the necessity to have interdomain traffic engineering working without interdomain collaboration. Finally, we will present our on-going work.

Different approaches have been developed to mitigate the state space explosion of model checking techniques. Among them, symbolic verification techniques use efficient representations such as BDDs to reason over sets of states rather than over individual states. Unfortunately, past experience has shown that these techniques do not work well for loosely-synchronized models. This paper presents a new algorithm and a new tool that combines BDD-based model checking with partial order reduction (POR) to allow the verification of models featuring asynchronous processes, with significant performance improvements over currently available tools. We start from the ImProviso algorithm (Lerda et al.) for computing reachable states, which combines POR and symbolic verification. We merge it with the FwdUntil method (Iwashita et al.) that supports verification of a subset of CTL. Our algorithm has been implemented in a prototype that is applicable to action-based models and logics such as process algebras and ACTL. Experimental results on a model of an industrial application show that our method can verify properties of a large industrial model which cannot be handled by conventional model checkers.

All PhD researchers at INGI will present their ongoing research in no more than 5 minutes (max. 1 to 3 slides per person). The idea of this session is to get to know each other to obtain a better vision on what research is currently going on in the department. Researchers that will present at this seminar are:

Researchers

1. Sébastien Combéfis
2. Bernard Lambeau (advisor: Axel van Lamsweerde)
3. Pham Quang Dung (advisor: Yves Deville) on "LS(Graph): A Local Search Framework for Constraint Optimization on graphs"
4. Christophe Damas (advisor: Axel van Lamsweerde)

Professors

1. Olivier Bonaventure
2. Gildas Avoine
3. Charles Pecheur
4. Yves Deville
5. Axel van Lamsweerde
6. Marc Lobelle
7. Kim Mens

All PhD researchers at INGI will present their ongoing research in no more than 5 minutes (max. 1 to 3 slides per person). The idea of this session is to get to know each other to obtain a better vision on what research is currently going on in the department. Researchers that will present at this seminar are:

1. Pierre Schaus (advisor: Yves Deville) is conducting research on constraint programming. The objective is to develop statistical constraints and propagators on common statistical objects like the mean, variance, median, percentile, histogram... and to illustrate the usefulness of these constraints on the Assembly Line Balancing Problem.
2. Jean-Noël Monette (advisor: Yves Deville) is conducting research on model-based synthesis of local search algorithms for scheduling problems.
3. Boriss Mejias (advisor: Peter Van Roy) is conducting research on self-management of large scale distributed systems.
4. Damien Saucez (advisor: Olivier Bonaventure) is working on traffic engineering and investigating ways for an ISP or network administrator to manage the traffic and P2P traffic in particular.
5. Sébastien Barré (advisor: Olivier Bonaventure) is conducting research on efficient use of IPv6 multihoming.
6. José Vander Meulen (advisor: Charles Pecheur) studies the symbolic model checking of LTS and mixed models and logics.
7. Virginie Van den Schriek (advisor: Olivier Bonaventure) is conducting research on improving iBGP path diversity of egress routers.
8. Sébastien Mouthuy (advisor: Yves Deville). Very Large-Scale Neighborhoods: towards generic abstractions.
9. Sergio Castro (adviosr: Kim Mens). Towards (semi)automatic support for inconsistency management in Intensional Views and constraints.
10. Damien Leroy (advisor: Oliver Bonaventure) conducts research on locator allocation and distribution in networks.
11. Helleputte Thibault (advisor: Pierre Dupont) on Microarray Data Analysis for Medical Prognosis.

Abstract: Over the past three decades, more and more research has been spent on understanding software evolution. However, the approaches developed so far rely on ad-hoc models, or on too specific meta-models, and thus, it is difficult to reuse or compare their results. We argue for the need of an explicit and generic meta-model that recognizes evolution as an explicit phenomenon and models it as a first class entity. Our solution is to encapsulate the evolution in the explicit notion of history as a sequence of versions, and to build a meta-model around these notions called Hismo. To show the usefulness of our meta-model we exercise its different characteristics by building several reverse engineering applications.

Bio: Tudor Girba attained the PhD degree in 2005 at the University of Berne, Switzerland, and since then he is working as senior researcher at the same university. His interests lie in the area of software engineering with focus on reengineering. He is one of the main architects and developers of Moose analysis platform, he developed the Hismo software evolution meta-model, he co-authored the Mondrian interactive visualization scripting engine, and he participated in the development of several other reverse engineering tools and models. He is the president of the Moose Association and he is member of the Executive Board of CHOOSE. He also offers consulting services in the area of reengineering and quality assurance.

Abstract: Understanding software systems is hampered by their sheer size and complexity. Software visualization encodes the data found in these systems into pictures and enables the human eye to interpret it. In this lecture we present the concepts of software visualization and we show several examples of how visualizations can help in understanding software systems. We will also complement the theory with practical demos using the Moose analysis platform (http://moose.unibe.ch).

Bio: Tudor Girba attained the PhD degree in 2005 at the University of Berne, Switzerland, and since then he is working as senior researcher at the same university. His interests lie in the area of software engineering with focus on reengineering. He is one of the main architects and developers of Moose analysis platform, he developed the Hismo software evolution meta-model, he co-authored the Mondrian interactive visualization scripting engine, and he participated in the development of several other reverse engineering tools and models. He is the president of the Moose Association and he is member of the Executive Board of CHOOSE. He also offers consulting services in the area of reengineering and quality assurance.

In this talk, I will present a summary of my current work on the synthesis of algorithms for Scheduling problems. The goal of this research is to allow a user to modelize its scheduling problem (and get a solution) without worrying about the search used to solve its problem. The model described by the user is analyzed to detect the properties of the problem and to generate the most suited algorithm. I will first present the modelling side of the developed system and how the analysis is performed. In a second part, I will talk about the actual local-search algorithms that are generated and used to solve some scheduling problems in the family of the well known Job-Shop. I'll try to make the talk understandable by people not familiar with Scheduling and/or Local-Search.

Comme chaque année, l'équipe sysadmin fait le point avec vous sur l'état de l'infrastructure informatique en INGI. Après une mise en jambes présentant l'état de nos différents services, nous exposerons nos réalisations de l'année écoulée, l'état d'avancement de nos projets en cours, ainsi que les évolutions futures envisagées pour notre système informatique. Nous espérons aussi que ce sera l'occasion d'échanger avec vous impressions, commentaires et désidératas à propos de notre outil de travail commun.

The application-oriented measurement platform (AOMP) offers a measurement service that provides network characteristics for applications so that the applications can optimize their operations according to the state of the Internet. A typical instance of the applications that can benefit from AOMP is an overlay network application, which reconstructs its overlay paths based on some metrics such as round-trip time between two nodes. In this talk, considering requirements imposed from the applications, we focus on the design choice of AOMP with usage scenarios. Moreover, we describe some challenges for deploying the platform in actual network environment.

Decentralization is gaining more and more importance in the construction of large scale distributed systems. By decentralizing the system, there is single point of failure, as in classic client-server architectures, and bottle-necks are highly reduced. Paradoxically, having no central point of control increases the complexity of the system. We propose to increase self-management to deal with such complexity.
We propose to build decentralized systems using structured overlay networks and distributed storage services. We need an overlay network that is not only self-organizing, but also fault-tolerant with self-healing properties. On top of this network, we build the distributed storage using transactions that rely on two properties: atomic commit and concurrency control. This means that transactions will not be affected by partial failures of the systems, and that concurrent transactions will not be interfered by partial results of other transactions. This IRM will be dedicated to introduce the maintenance of a structured overlay network, and the mechanisms to provide the above mentioned properties needed for transactions in a decentralised system.

The possibility to evolve software at runtime is an inevitable necessity in certain scenarios. Not only does it greatly increase programmer productivity at development time, it also widens the range of deployment opportunities by enabling software updates while keeping system downtimes to a minimum.

Most approaches for enabling dynamic software evolution are based on application-wide globally visible effects (Common Lisp's "change-class", Smalltalk's "become:", Java's "HotSwap", and so on). They all share a common problem: As soon as a dynamic update occurs, it is more or less immediately visible in all currently executing threads. This may render the execution state of some threads inconsistent because they rely on program definitions that were in effect before the change, while other threads may preferably already use the new definitions. Technical solutions for this problem have been suggested, but the conceptual burden of coordinating software updates remains exceptionally high.

Since three years, we have been working on a new programming paradigm called Context-oriented Programming (COP). The main idea of COP is to enable expressing context-dependent program behavior in a concise way, which is already finding its way into several industrial applications worldwide. In this talk, however, we present a different perspective on our ideas and show how COP can support dynamic software evolution while avoiding the typical problems associated with updates of program definitions at runtime. I will show examples of the use of COP for dynamic software evolution, illustrate how COP can help to avoid the typical problems of dynamic software evolution, sketch some implementation strategies for COP language extensions, and discuss implications for dynamically evolvable software architectures.

Pascal Costanza has a Ph.D. degree from the University of Bonn, Germany, and works as a research assistant at the Programming Technology Lab of the Vrije Universiteit Brussel, Belgium. His past involvements include specification and implementation of the languages Gilgul and Lava, and the design and application of the JMangler framework for load-time transformation of Java class files. He has also implemented ContextL, the first programming language extension for Context-oriented Programming based on Common Lisp, and aspect-oriented extensions for Common Lisp.