What are the main mechanisms involved in an asthma crisis? How does a curative aerosol or the dust pollution deposit in the airways? How emphysema or fibrosis affects the supply of air? What is the function of the surfactant that covers lung alveoli? … From the medical, environmental and societal point of view, this kind of questions raises many challenging issues we would like to contribute to answer. Indeed we think that the mathematical approach of the problem, in interaction with lung specialists, could provide new elements and improve the understanding on the topic. In particular, numerical simulations may accurately reproduce the behaviour of the airflow in the bronchial tree and they may help the clinician to identify the pathology and cure it. The present research program thus focuses on the mathematical, numerical and computational modelling of the respiratory system. The complex fractal geometry of the airway tree makes the airflow simulation on the whole tree unreachable, not to mention the global simulation of the interaction between an aerosol and the air. Consequently, it is necessary to find new efficient strategies, including simple but realistic models. Depending on the type of applications we want to study (air transport, particle deposition, mechanical behaviour of the parenchyma, airway closure), we intend to find such adequate and realistic models by taking into account each physical phenomenon and coupling them one to each other, or deriving boundary conditions (or reduced model) to describe the part of the lung which is not under focus. Then, for each of these models, many questions may be addressed, such as the wellposedness of the problems, their mathematical properties (regularity of the solution, long time behaviour, controllability of the system), the design of accurate numerical algorithms, the performing of efficient computational simulations. At each stage, we shall discuss our results with experts from other fields (fluid and structure mechanics as well as physiology and medicine). In particular, our computations will be compared to experimental results or clinical data and it will be used as a feedback until good agreement is reached and they can serve as a prediction tool.

Pervasive Computing (PerCom) envisions the unobtrusive diffusion of computing and networking resources in physical environments, enabling users to access information and computation anywhere and at anytime, in a user-centric way. Ambient Intelligence (AmI) is a strategic domain for the EC that complements PerCom with a more everyday user- or consumer-oriented application of computation integrating into physical environments. Service-Oriented Computing (SOC) is a recent computing paradigm which seems particularly appropriate to PerCom and AmI. Therein, networked devices and their hosted applications are abstracted as services delivered and consumed on demand. The interest in SOC (e.g., using Web Services – WS) has vertically increased in the last five years as it can provide the answer to the integration of heterogeneous devices, software platforms, applications and, finally, whole systems, whether on the Web, on wireless networks or on enterprise intranets. Nevertheless, the SOC paradigm alone cannot guarantee valid (automatic) service composition in the open pervasive environment. Issues include: (i) interaction between services is based on their syntactic description, for which common understanding is hardly achievable in open environments; (ii) service behaviour is often assumed to be simple, i.e., reduced to stateless service invocations, or assumed to follow a standard interaction model, e.g., client-server; and (iii) non-functional properties of service compositions, such as quality of service (QoS) and context, are not taken into account, which makes resulting user's satisfaction in AmI highly uncertain. The objectives of the proposed research project comprise the study, analysis and elaboration of a comprehensive approach to service composition in pervasive environments. We explicitly associate the research issues identified above with three description levels for service interfaces: behavioural level, non-functional level and semantic level. These three levels enrich the standard description level for service interfaces, which is the service signature level. We specifically aim at supporting all three advanced description levels and their integration towards a thorough approach to service composition. Two axes are further identified in the research. The first axis concerns the foundations of service composition in terms of underlying formal models and related algorithms for: service interface specification; service discovery based on a required interface specification; service composition and adaptation, the latter when composed services do not perfectly match; service execution semantics; and composite service reconfiguration, when conditions change during service execution. More specifically, we propose to base interface descriptions on enrichments of formal models such as Labelled Transition Systems (yielding, e.g., Symbolic Transition Systems) or Process Algebras (yielding, e.g., Temporal Process Algebras). We have shown in recent work that integration of formal models is possible on top of a strong formal basis. We propose to exploit and extend further these results towards the integration of interface description levels both from a model viewpoint – which will enable integrated model-based service composition at different interface description levels – and from an algorithmic viewpoint – which will enable integration of the individual service composition algorithms related to each interface description level. We have in previous work produced base results regarding the latter viewpoint. Model-based composition techniques will be addressed first in order to promote formal and language-independent solutions. In a second step, model transformation into selected service languages or extensions will be tackled to associate models with existing service language standards. Further on, we will address assessment of the proposed solutions through the development of a prototype tool for model-based service composition. The special features of the pervasive environment will be taken into account all through our approach and always in view of our second research axis. This second axis concerns the application of the elaborated first axis outcomes to the runtime pervasive environment, tackling issues such as: efficiency and performance of algorithms for the interactive pervasive environment; resource consumption on resource-constrained portable devices; monitoring mechanisms for detecting change of conditions and for triggering composite service reconfiguration; runtime mechanisms for ensuring QoS in the face of change.

The goal of this project is to study so-called multiple-impact laws, with and without friction, in order on one hand to improve the capabilities of numerical methods, on the other hand to better understand the dynapics of nonsmooth multibody mechanical systems. The two aspects (theory and numerics) are closely linked oe to each other, since numerical simulations have become a mandatory path to better understand the dynamics of complex systems. Thisi is true in industry with virtual prototyping, and in the academic field (granualr media, robotic systems). Multiple impacts (several impacts at the same time in the system) are a crucial phenomenon and approximate modelling that does not correctly take into account 1) the local dissipation at the contact points, 2) the global effects of energy dispersion, which are mainly due to wave effects that travel through the system, are not acceptable. The goal of this project is to design multiple restitution laws which correctly model these two aspects, and which depends on identifiable parameters, and which can be reliably simulated (no stiff equations). Moreover such a restitution law should be generic enough to correctly predict the dynamics of significantly different sysems (for instance a chain of iron balls, or a chain of tennis balls). The preliminary results obtained in C. Liu, Z. Zhao, B. Brogliato, Theoretical analysis and numerical algorithm for frictionless multiple impacts in multibody systems, INRIA research Report, available at http://hal.inria.fr are quite promissing. A third aspect of the project is the design and construction of experimental setups, which will allow us to validate the theoretical and numerical results (and in turn which are indispensable to circumvent the main physical effects of a multiple impact). The targetted applications are granular media and kinematic chains. The theoretical works will be led by the two partners in close cooperation. The numerical aspects will be led by the French partner, which has a strong experience in software platform development (with the SICONOS platform). The experiments will be designed by the Chinese partner, which has a strong experience with the CAST. The three sides of the project (theory, numerics, experiments) will be led in parallel for obvious reasons.

The recent succes of collision attacks on cryptographic hash functions have created some turmoil in the cryptologic research community. In particular, collisions for MD5 and SHA-0 have been exhibited, and some costly but feasible attacks on SHA-1 have been described. This represent a potential threat for some cryptographic applications, as digital signatures. We propose two main topics for our project. 1) Understand the best attacks, in particular those presented by the Chinese researcher Wang. Actually, she provides some equations that allow the construction of collisions. It is possible to check the validity of those equations, but the way they are obtained is not very clear and deserves to be carefully examined and may hopefully help to define design criteria. 2) Find new design techniques for secure hash functions. In particular, we propose to extend some recent works on hash functions with security reduction. By combining those functions using constructions `a la Merkle-Damgard, we hope to produce fast cryptographic hash functions with a proof of security.