RENAULT Eric

From the fifth generation onwards, mobile networks will have to support an exponential growth in the number of connected devices of various types, this being one of the pillars of a global strategy of accelerated digitization. In addition to this growth in connectivity, these networks will also have to support and deliver various services for new industries with heterogeneous requirements. 5G network designers and developers are then forced to provide new solutions and optimize existing ones to contain the growing bandwidth demands and higher quality of experience (QoE) expectations. These networks also need to be highly customizable to fit various use cases and highly automated to shorten time-to-market. The expected characteristics of 5G networks have prompted mobile network providers to radically change the way they design and develop their solutions, adopting a strategy where software-based solutions are preferred. As the mobile networking domain and the rest of the IT world converge, mobile network providers can benefit from practices and tools at the forefront of burgeoning software and cloud computing ecosystems. Software network functions would allow these providers to have the levels of programmability and reconfigurability they need to cope with such rapid evolution of mobile connectivity. This thesis aims to provide some optimizations of different parts of 5G networks and how they are deployed and managed, with the hope that this will help solve some of the problems facing mobile network designers. This thesis proposes solutions to specific problems related to physical layer processing in 5G networks for interference mitigation, as well as generic problems related to network automation and customization. In this thesis, we have built a platform, which is used to create end-to-end mobile networks, composed of a radio access network (RAN) platform, network core and orchestration, using the concepts and tools of the meta platform. The first part addresses the issue of inter-cell interference, which is likely to be a handicap with the expected densification of antennas in 5G networks. We propose a solution to mitigate the effects of this interference for transmissions from mobiles to base stations. This solution is based on the multi-receiver detection (JD) technique. It meets the architectural, functional and technical requirements for the integration of JD in practical networks. We integrate the JD solution into a RAN platform in the second part and extend this platform with other functionalities. We take the same approach in the third part of this thesis to provide a solution to automate core network deployment and lifecycle management in a network functions virtualization (NFV) environment and create a reusable core network platform orchestrated by open network automation platform (ONAP).

No summary available.

No summary available.

No summary available.

Indoor Localization has attracted interest in both academia and industry for its wide range of applications. In this paper, we propose an indoor localization solution based on Channel State Information (CSI). CSI is a fine-grain measure of the effect of the channel on the transmitted signal. It is computed for each subcarrier and each antenna in the Multiple-Input-Multiple-Output (MIMO) antenna case. It is also becoming a trend for indoor position fingerprinting. By using a K-nearest neighbor learning method a highly accurate indoor positioning is achieved. The input feature is the magnitude component of CSI which is preprocessed to reduce noise and allow for a quicker search. The euclidean distance between CSI is the criteria chosen for measuring the closeness between samples. The method is applied to a CSI dataset estimated at an 8 × 2 MIMO antenna that is published by the organizers of the Communication Theory Workshop Indoor Positioning Competition. The proposed method is compared with three other methods all based on deep learning approaches and tested with the same dataset. The K-nearest neighbor method presented in this paper achieves a Mean Square Error (MSE) of 2.4 cm which outperforms its counterparts.

Localization is the process of estimating the position of an entity in a local or global coordinate system. Localization applications are widely distributed in different contexts. In events, tracking participants can save lives during crises. In healthcare, elderly people can be tracked to meet their needs in critical situations such as falls. In warehouses, robots transferring products from one place to another require precise knowledge of its position, the position of the products as well as of other robots. In an industrial context, localization is essential to realize automated processes that are flexible enough to be reconfigured to various missions. Localization is considered a topic of great interest in both industry and academia, especially with the advent of 5G with its "Enhanced Mobile Broadband (eMBB)" which is expected to reach 10 Gbps, "Ultra-Reliable Low-Latency Communication (URLLC)" which is less than one millisecond and "massive Machine-Type Communication (mMTC)" allowing to connect about 1 million devices per kilometer.In this work, we focus on two main types of localization. distance-based localization between devices and fingerprint-based localization. In distance-based localization, a network of devices with a maximum communication distance estimates distance values from their neighbors. These distances along with the knowledge of the positions of some nodes are used to locate other nodes in the network using a triangulation-based solution. The proposed method is able to locate about 90% of the nodes in a network with an average degree of 10.In fingerprint-based localization, the responses of wireless channels are used to estimate the position of a transmitter communicating with a MIMO antenna. In this work, we apply classical learning techniques (K-nearest neighbors) and deep learning techniques (Multi-Layer Perceptron Neural Network and Convolutional Neural Networks) to localize a transmitter in indoor and outdoor settings. Our work won the first prize in the positioning competition prepared by IEEE Communication Theory Workshop among 8 teams from highly reputable universities around the world by obtaining a mean square error of 2.3 cm.

Data Centers (DCs) need to periodically configure their servers in order to meet user demands. Since appropriate proactive management to meet demands reduces the cost, either by improving Quality of Service (QoS) or saving energy, there is a great interest in studying different proactive strategies based on predictions of the energy used to serve CPU and memory requests. The amount of savings that can be achieved depends not only on the selected proactive strategy but also on user-demand statistics and the predictors used. Despite its importance, it is difficult to find theoretical studies that quantify the savings that can be made, due to the problem complexity. A proactive DC management strategy is presented together with its upper bound of energy cost savings obtained with respect to a purely reactive management. Using this method together with records of the recent past, it is possible to quantify the efficiency of different predictors. Both linear and nonlinear predictors are studied, using a Google data set collected over 29 days, to evaluate the benefits that can be obtained with these two predictors.

No summary available.

Mobile networks currently use an imperative approach to network service delivery and service lifecycle management. The technology leaps that come with 5G will attract millions of new users and huge volumes of data. Network infrastructure will reach such complexity that imperative mode management will not be able to keep up with the expected increase in service demands. Software Defined Networking (SDN) and Network Function Virtualization (NFV) technologies are paving the way for programmability, flexibility and scalability of mobile networks. Both technologies offer a significant advantage to Network Operators (NOs) in terms of network management and service delivery, and expand their market to third-party providers such as Virtual Network Operators (VNOs) and Over-The-Top (OTT) application providers. However, these technologies still rely on imperative approaches to network service management and delivery. A declarative approach to network service management is required to manage their network growth in a seamless manner, which is provided by an intent-based networking (IBN) approach. IBN involves organizing and abstracting complex sets of network management and configuration instructions to expose them to network tenants in the form of a simple and unambiguous service request called an Intent. The Intent describes WHAT is requested while the network manages HOW to respond. This thesis proposes an Intent-based processing framework for vertical market request processing. The study focuses on the provisioning of application-dedicated 5G network slices. The framework helps both operators and network tenants to express their intent in a 4th generation language close to human language and in transformation language (source-to-source).

No summary available.

No summary available.

Location-based routing protocols for vehicular ad hoc networks (VANETs) use location information to determine routing decisions. This information is provided by a location service that is queried by nodes in order to properly forward packets to communication partners. This paper presents the semiflooding location service, a proactive flooding-based location service that drastically reduces the number of update packets sent over the network compared to traditional flooding-based location services. This goal is achieved by each node partially forwarding location information. We present both deterministic and probabilistic approaches for this algorithm, which remains very simple. A mathematical model is proposed to show the effectiveness of this solution. The cases of homogeneous 1D, 2D, and 3D networks were studied for both deterministic and probabilistic forwarding decisions. We compare our algorithm with simple flooding and with the multipoint-relay (MPR) flooding of the optimized-link-state-routing (OLSR) protocol, and we show that our algorithm, despite being very simple, has excellent scalability properties. The mean number of generated messages ranges with the mean number of the neighbors of one random network node.

Indoor localization has attracted much attention due to its indispensable applications e.g. autonomous driving, Internet-Of-Things (IOT), and routing, etc. Received Signal Strength Indicator (RSSI) was used intensively to achieve localization. However, due to its temporal instability, the focus has shifted towards the use of Channel State Information (CSI) aka channel response. In this paper, we propose a deep learning solution for the indoor localization problem using the CSI of a 2 × 8 Multiple Input Multiple Output (MIMO) antenna. The variation of the magnitude component of the CSI is chosen as the input for a multi- layer Perceptron (MLP) neural network. Data augmentation is used to improve the learning process. Finally, various MLP neural networks are constructed using different portions of the training set and different hy- perparameters. An ensemble neural network technique is then used to process the predictions of the MLPs in order to enhance the position estimation. Our method is compared with two other deep learning so- lutions. one that uses Convolutional Neural Network (CNN), and the other uses MLP. The proposed method yields higher accuracy than its counterparts.

No summary available.

Data Center (DC) management aims at promptly serving user requests while minimizing the energy consumed. This is achieved by turning off unnecessary servers to save energy and adapting the number of servers that are on to the time-varying and heterogeneous user requests. A great change in the number of servers on leads to a considerable management effort, also called control effort in the literature, which should be reduced as much as possible. Since feedback control can improve the performance of computing systems and networks, we propose to use it to achieve this dynamic capacity provisioning of the DC. In order to design this feedback control, first, we developed a dynamic model of the DC. The purpose of this paper is to design a feedback control strategy based on the DC model, able to optimize i) the Quality of Service, ii) the energy consumed and iii) the management effort. A simple Reactive open-loop Control which provides an amount of energy equal to the amount requested in the previous time interval is considered as a benchmark for comparison. Second, two feedback controls based on the balance equations of the DC are studied, namely i) Reactive Feedback Control providing an amount of energy equal to that provided by the reactive open-loop control but adding the accumulated demand that has not yet been served, and ii) Model Predictive Control optimizing a constrained cost that weights the management effort and the prediction error. Reactive Control, Reactive Feedback Control and Model Predictive Control are compared in terms of energy consumed, energy error and management effort. Quantitative results of the comparative performance evaluation are given, based on a data set collected from a real DC. INDEX TERMS Data center management, energy efficiency, quality of service, dynamic capacity provi-sioning, reactive control, reactive feedback control, model predictive control.

Due to the emerging demand for IoT applications, indoor positioning became an invaluable task. We propose a novel lightweight deep learning solution to the indoor positioning problem based on noise and dimensionality reduction of MIMO Channel State Information (CSI). Based on preliminary data analysis, the magnitude of the CSI is selected as the input feature for a Multilayer Perceptron (MLP) neural network. Polynomial regression is then applied to batches of data points to filter noise and reduce input dimensionality by a factor of 14. The MLP’s hyperparameters are empirically tuned to achieve the highest accuracy. The proposed solution is compared with a state-of-the-art method presented by the authors who designed the MIMO antenna that is used to generate the dataset. Our method yields a mean error which is 8 times less than that of its counterpart. We conclude that the arithmetic mean and standard deviation misrepresent the results since the errors follow a log- normal distribution. The mean of the log error distribution of our method translates to a mean error as low as 1.5 cm.

Data Centers (DCs) need to manage their servers periodically to meet user demand efficiently. Since the cost of the energy employed to serve the user demand is lower when DC settings (e.g. number of active servers) are done a priori (proactively), there is a great interest in studying different proactive strategies based on predictions of requests. The amount of savings in energy cost that can be achieved depends not only on the selected proactive strategy but also on the statistics of the demand and the predictors used. Despite its importance, due to the complexity of the problem it is difficult to find studies that quantify the savings that can be obtained. The main contribution of this paper is to propose a generic methodology to quantify the possible cost reduction using proactive management based on predictions. Thus, using this method together with past data it is possible to quantify the efficiency of different predictors as well as optimize proactive strategies. In this paper, the cost reduction is evaluated using both ARMA (Auto Regressive Moving Average) and LV (Last Value) predictors. We then apply this methodology to the Google dataset collected over a period of 29 days to evaluate the benefit that can be obtained with those two predictors in the considered DC.

Indoor localization has attracted much attention due to its indispensable applications e.g. autonomous driving, Internet-Of-Things (IOT), and routing, etc. Received Signal Strength Indicator (RSSI) was used intensively to achieve localization. However, due to its temporal instability, the focus has shifted towards the use of Channel State Information (CSI) aka channel response. In this paper, we propose a deep learning solution for the indoor localization problem using the CSI of a 2 × 8 Multiple Input Multiple Output (MIMO) antenna. The variation of the magnitude component of the CSI is chosen as the input for a multi- layer Perceptron (MLP) neural network. Data augmentation is used to improve the learning process. Finally, various MLP neural networks are constructed using different portions of the training set and different hy- perparameters. An ensemble neural network technique is then used to process the predictions of the MLPs in order to enhance the position estimation. Our method is compared with two other deep learning so- lutions. one that uses Convolutional Neural Network (CNN), and the other uses MLP. The proposed method yields higher accuracy than its counterparts.

Due to the emerging demand for IoT applications, indoor positioning became an invaluable task. We propose a novel lightweight deep learning solution to the indoor positioning problem based on noise and dimensionality reduction of MIMO Channel State Information (CSI). Based on preliminary data analysis, the magnitude of the CSI is selected as the input feature for a Multilayer Perceptron (MLP) neural network. Polynomial regression is then applied to batches of data points to filter noise and reduce input dimensionality by a factor of 14. The MLP’s hyperparameters are empirically tuned to achieve the highest accuracy. The proposed solution is compared with a state-of-the-art method presented by the authors who designed the MIMO antenna that is used to generate the dataset. Our method yields a mean error which is 8 times less than that of its counterpart. We conclude that the arithmetic mean and standard deviation misrepresent the results since the errors follow a log- normal distribution. The mean of the log error distribution of our method translates to a mean error as low as 1.5 cm.

This book constitutes the thoroughly refereed proceedings of the First International Conference on Machine Learning for Networking, MLN 2018, held in Paris, France, in November 2018. The 22 revised full papers included in the volume were carefully reviewed and selected from 48 submissions. They present new trends in the following topics: Deep and reinforcement learning. Pattern recognition and classification for networks. Machine learning for network slicing optimization, 5G system, user behavior prediction, multimedia, IoT, security and protection. Optimization and new innovative machine learning methods. Performance analysis of machine learning algorithms. Experimental evaluations of machine learning. Data mining in heterogeneous networks. Distributed and decentralized machine learning algorithms. Intelligent cloud-support communications, resource allocation, energy-aware/green communications, software defined networks, cooperative networks, positioning and navigation systems, wireless communications, wireless sensor networks, underwater sensor networks.

No summary available.

Identification of road anomaly not only helps drivers to reduce the risk, but also support for road maintenance. Arguably, with the popularity of smartphones including multiple sensors, many road anomaly detection systems using mobile phones have been proposed. This paper aims at analyzing a number of typical road anomaly detection methods in terms of resource requirements, energy consumption, fitness conditions. From these measurements, we suggest some improvement directions to build road anomaly detection algorithms appropriate for smartphones.

Localization in Mobile Ad-hoc Networks (MANETs) and Wireless Sensor Networks (WSNs) is an issue of great interest, especially in applications such as the IoT and VANETs. We propose a solution that overcomes two limiting characteristics of these types of networks. The first is the high cost of nodes with a location sensor (such as GPS) which we will refer to as anchor nodes. The second is the low computational capability of nodes in the network. The proposed algorithm addresses two issues. self-localization where each non-anchor node should discover its own position, and global localization where a node establishes knowledge of the position of all the nodes in the network. We address the problem as a graph where vertices are nodes in the network and edges indicate connectivity between nodes. The weights of edges represent the Euclidean distance between the nodes. Given a graph with at least three anchor nodes and knowing the maximum communication range for each node, we are able to localize nodes using fairly simple computations in a moderately dense graph.

The rapid deployment of new infrastructures based on network virtualization and Cloud computing triggers new applications and services that in turn generate new constraints such as security and/or mobility. The International Conference on Mobile, Secure and Programmable Networking (MSPN) aimed at providing a top forum for researchers and practitioners to present and discuss new trends in networking infrastructures, security, services and applications while focusing on virtualization and Cloud computing for networks, network programming, Software Defined Networks (SDN) and their security. In 2018, MSPN was hosted by CNAM Paris, one of the oldest teaching center in Paris. The call for papers resulted in a total of 52 submissions from all around the world. Every submission was assigned to at least three members of the program committee for review. The program committee decided to accept 27 papers. The accepted papers are from: Algeria, Argentina, Benin, China, Colombia, France, Germany, India, Morocco, Norway, The Netherlands, Tunisia, The United Kingdom and Vietnam. Two intriguing keynotes from Professor Ruben Milocco of the University of Comahue, Argentina, and Professor Jean-Claude Belfiore, Head of the Communications Science Department at Huawei Technologies, France, completed the technical program. We would like to thank all who contributed to the success of this conference, in particular the members of the program committee and the reviewers for carefully reviewing the contributions and selecting a high-quality program. Our special thanks go to the members of the organizing committee for their greatful help. We hope that all participants enjoyed this successful conference, made a lot of new contacts, engaged in fruitful discussions, and had a pleasant stay in Paris, France.

Location services for ad-hoc networks are of indispensable value for a wide range of applications, such as the Internet of Things (IoT) and vehicular ad-hoc networks (VANETs). Each context requires a solution that addresses the specific needs of the application. For instance, IoT sensor nodes have resource constraints (i.e., computational capabilities), and so a localization service should be highly efficient to conserve the lifespan of these nodes. We propose an optimized energy-aware and low computational solution, requiring 3-GPS equipped nodes (anchor nodes) in the network. Moreover, the computations are lightweight and can be implemented distributively among nodes. Knowing the maximum range of communication for all nodes and distances between 1-hop neighbors, each node localizes itself and shares its location with the network in an efficient manner. We simulate our proposed algorithm in a NS-3 simulator, and compare our solution with state-of-the-art methods. Our method is capable of localizing more nodes (≈ 90% of nodes in a network with an average degree ≈ 10).

Wireless and mobile networks have developed enormously in recent years. Far from being reserved for industrialized countries, these networks requiring a limited fixed infrastructure have also imposed themselves in emerging and developing countries. Indeed, with a relatively low structural investment compared to the one necessary for the implementation of a wireline network, these networks allow operators to offer a very wide coverage of the territory, with a cost of access to the network (price of the telephone and the communications) completely acceptable for the users. Therefore, it is not surprising that today, in most countries, the number of wireless phones is much higher than the number of fixed phones. This large number of terminals spread all over the world is an invaluable reservoir of information, of which only a tiny part is currently exploited. Indeed, by combining the position of a cell phone and its speed of movement, it becomes possible to deduce the quality of roads or road traffic. On another note, by integrating a thermometer and/or a hygrometer in each terminal, which on a large scale would imply a derisory unit cost, these terminals could serve as a relay for a more reliable local weather forecast. In this context, the objective of this thesis is to study and analyze the opportunities offered by the use of data from mobile terminals, to propose original solutions for the processing of these large masses of data, insisting on the optimizations (fusion, aggregation, etc.) that can be realized in an intermediate way in the context of their transport to the storage and processing center(s), and eventually to identify the data not available today on these terminals but that could have a strong impact in the years to come. A prototype presenting a typical example of use will allow to validate the different approaches.

This book constitutes the thoroughly refereed post-conference proceedings of the 5th International Conference on Mobile, Secure and Programmable Networking, held in Mohammedia, Morocco, in April 2019. The 23 papers presented in this volume were carefully reviewed and selected from 48 submissions. They discuss new trends in networking infrastructures, security, services and applications while focusing on virtualization and cloud computing for networks, network programming, software defined networks (SDN) and their security.

No summary available.

No summary available.

Data Centers (DCs) need to manage their servers periodically to meet user demand efficiently. Since the cost of the energy employed to serve the user demand is lower when DC settings (e.g. number of active servers) are done a priori (proactively), there is a great interest in studying different proactive strategies based on predictions of requests. The amount of savings in energy cost that can be achieved depends not only on the selected proactive strategy but also on the statistics of the demand and the predictors used. Despite its importance, due to the complexity of the problem it is difficult to find studies that quantify the savings that can be obtained. The main contribution of this paper is to propose a generic methodology to quantify the possible cost reduction using proactive management based on predictions. Thus, using this method together with past data it is possible to quantify the efficiency of different predictors as well as optimize proactive strategies. In this paper, the cost reduction is evaluated using both ARMA (Auto Regressive Moving Average) and LV (Last Value) predictors. We then apply this methodology to the Google dataset collected over a period of 29 days to evaluate the benefit that can be obtained with those two predictors in the considered DC.

OpenMP and MPI have become the standard tools for developing parallel programs on a shared memory and distributed memory architecture respectively. Compared to MPI, OpenMP is easier to use. This is due to the fact that OpenMP automatically generates the parallel code and synchronizes the results using directives, clauses and execution functions, while MPI requires programmers to do this work manually. As a result, efforts have been made to port OpenMP to distributed memory architectures. However, excluding CAPE, no solution satisfies both of the following requirements: 1) to be fully compliant with the OpenMP standard and 2) to have high performance. CAPE (Checkpointing-Aided Parallel Execution) is a framework that automatically translates and provides execution functions to run an OpenMP program on a distributed memory architecture based on checkpointing techniques. In order to execute an OpenMP program on a distributed memory system, CAPE uses a set of templates to translate the OpenMP source code into CAPE source code, and then the CAPE source code is compiled by a conventional C/C++ compiler. Basically, the idea of CAPE is that the program first runs on a set of nodes in the system, with each node functioning as a process. Whenever the program encounters a parallel section, the master distributes tasks to the slave processes using discontinuous incremental checkpoints (DICKPT). After sending the checkpoints, the master waits for the results returned by the slaves. The next step at the master level is to receive and merge the results of the checkpoints before injecting them into its memory. The slave nodes receive the various checkpoints and then inject them into their memory to perform the assigned work. The result is then sent back to the master using DICKPT. At the end of the parallel region, the master sends the checkpoint result to each slave to synchronize the program memory space. In some experiments, CAPE has shown high performance on distributed memory systems and is a viable solution fully compatible with OpenMP. However, CAPE is still in the development phase, as its checkpoints and execution model need to be optimized to improve performance, capacity and reliability. This thesis aims to present the proposed approaches to optimize and improve the capacity of checkpoints, design and implement a new execution model, and improve the capacity of CAPE. First, we proposed an arithmetic on checkpoints that models their data structure and its operations. This modeling helps to optimize their size and reduce the time required for merging, while improving their capacity. Second, we developed TICKPT (Time-Stamp Incremental Checkpointing), an implementation of checkpoint arithmetic. TICKPT is an improvement of DICKPT, it added time-stamp to the checkpoints to identify their order. Analysis and comparative experiments show that TICKPT is not only smaller, but also has less impact on program performance. Third, we designed and implemented a new execution model and prototypes for CAPE based on TICKPT. The new execution model allows CAPE to use resources efficiently, avoid the risk of bottlenecks and satisfy the requirement of Bernstein's conditions. In the end, these approaches significantly improve CAPE's performance, capabilities and reliability. The data sharing implemented on CAPE and based on arithmetic on checkpoints is open and based on TICKPT. This also demonstrates the right direction we have taken and makes CAPE more complete.

Several big data-driven applications are currently carried out in collaboration using distributed infrastructure. These data-driven applications usually deal with experiments at massive scale. Data generated by such experiments are huge and stored at multiple geographic locations for reuse. Workflow systems, composed of jobs using collaborative task-based models, present new dependency and data exchange needs. This gives rise to new issues when selecting distributed data and storage resources so that the execution of applications is on time, and resource usage-cost-efficient. In this paper, we present an efficient data placement approach to improve the performance of workflow processing in distributed data centres. The proposed approach involves two types of data: splittable and unsplittable intermediate data. Moreover, we place intermediate data by considering not only their source location but also their dependencies. The main objective is to minimise the total storage cost, including the effort for transferring, storing, and moving that data according to the applications needs. We first propose an exact algorithm which takes into account the intra-job dependencies, and we show that the optimal fractional intermediate data placement problem is NP-hard. To solve the problem of unsplittable intermediate data placement, we propose a greedy heuristic algorithm based on a network flow optimisation framework. The experimental results show that the performance of our approach is very promising. We also show that even with divergent conditions, the cost ratio of the heuristic approach is close to the optimal solution.

Traces collected from an operational Google data center over 29 days represent a very rich and useful source of information for understanding the main features of a data center. In this paper, we characterize the strong heterogeneity of jobs and the medium heterogeneity of machine configurations. We analyze the off-periods of machines. We study the distribution of jobs per category, per scheduling class, per priority and per number of tasks. The distribution of job execution durations shows a high disparity, as does the job waiting time before being scheduled. The resource requests in terms of CPU and memory are also analyzed. The distribution of these parameter values is very useful to develop accurate models and algorithms for resource allocation in data centers.

Checkpointing-Aided Parallel Execution (CAPE) is a framework that is based on checkpointing technique and serves to automatically translates and execute OpenMP programs on distributed-memory architectures. In some comparisons with MPI, CAPE have demonstrated high-performance and the potential for fully compatibility with OpenMP on distributed-memory systems. However, it should be continued to improve the performance, flexibility, portability and capability. This paper presents the new execution model for CAPE that improves its performance and makes CAPE even more flexible.

CAPE (Checkpointing-Aided Parallel Execution) is a framework that automatically translates and executes OpenMP on distributed-memory architectures based on checkpoint technique. In some experiments, this approach shows high-performance on distributed-memory system. However, it has not been fully developed yet. This paper presents an implementation of OpenMP data-sharing on CAPE that improves the capability, reduces checkpoint size and makes CAPE even more performance.

CAPE, which stands for Checkpointing-Aided Parallel Execution,is a checkpoint-based approach to automatically translate and execute OpenMP programs on distributed-memory architectures. This approach demonstrates high-performance and complete compatibility with OpenMP on distributed-memory systems. In CAPE, checkpointing is one of the main factors acted on the performance of the system. This is shown over two versions of CAPE. The first version based on complete checkpoints is too slow as compared to the second version based on Discontinuous Incremental Checkpointing. This paper presents an improvement of Discontinuous Incremental Checkpointing, and a new execution model for CAPE using new techniques of checkpointing. It contributes to improve the performance and make CAPE even more flexible.

Data collected from an operational Google data center during 29 days represent a very rich and very useful source of information for understanding the main features of a data center. In this paper, we highlight the strong heterogeneity of jobs. The distribution of job execution duration shows a high disparity, as well as the job waiting time before being scheduled. The resource requests in terms of CPU and memory are also analyzed. The knowledge of all these features is needed to design models of jobs, machines and resource requests that are representative of a real data center.

CAPE (Checkpointing-Aide Parallel Execution) is an approach tried to port OpenMP programs on distributed memory architectures like Cluster, Grid or Cloud systems. It provides a set of prototypes and functions to translate automatically and execute OpenMP program on distributed memory systems based on the checkpointing techniques. This solution has shown that it has achieved high performance and complete compatibility with OpenMP. However, it is in research and development stage, so there are many functions that need to be added, some techniques and models need to be improved. This paper presents approaches and techniques that have been applied and will be applied to optimize checkpoints and execution model of CAPE.

Collaborative cloud storage environment, which share resources of multiple geographically distributed datacenters owned by different providers enable scientific workflow from different locations to process large scale big intermediate data through the Internet. Distributed datacenters are federated and each member can collaborate with each other to efficiently share and process the intermediate data from distributed workflow instances. This paper focuses on the storage cost minimization of intermediate data placement in federated cloud datacenters. Through collaborative and federation mechanisms, we propose an exact federation data placement algorithm based on integer linear programming model (ILP) to assist multiple datacenters hosting intermediate data files generated from a scientific workflow. Under the constraints of the problem, the proposed algorithm finds an optimal intermediate data placement with a cost saving over the federated cloud datacenters, taking into account scientific user requirements, data dependency and size. Experimental results show the cost-efficiency of the proposed cloud storage federation algorithm.

Workflows are typical systems dealing with big data. These systems are deployed in geo-distributed locations to leverage existing cloud infrastructures and perform large-scale experiments. The data generated by such experiments is huge and stored in multiple locations for reuse. Indeed, workflow systems are composed of collaborative tasks, presenting new requirements in terms of dependency and intermediate data exchange for their processing. This leads to new problems in selecting distributed data and storage resources so that task or job execution is timely and resource utilization is cost-effective. Therefore, this thesis addresses the problem of managing data hosted in cloud data centers by considering the requirements of the workflow systems that generate them. To this end, the first problem addressed in this thesis deals with the intermediate data access behavior of tasks that are executed in a MapReduce-Hadoop cluster. This approach develops and explores the Markov model that uses the spatial location of blocks and analyzes the sequentiality of spill files through a prediction model. Second, this thesis addresses the problem of placing intermediate data in federated cloud storage by minimizing the storage cost. Through federation mechanisms, we propose an exact ILP algorithm to support multiple cloud data centers hosting dependency data by considering each pair of files. Finally, a more generic problem is addressed involving two variants of the placement problem related to divisible and integer dependencies. The main objective is to minimize the operational cost based on the requirements of inter and intra-job dependencies.

The rapid deployment of new infrastructures based on network virtualization and cloud computing triggers new applications and services that in turn generate new constraints such as security and/or mobility. The International Conference on Mobile, Secure and Programmable Networking aims at providing a top forum for researchers and practitioners to present the future networking infrastructures, services and applications and their security. MSPN 2017 was hosted by CNAM (Conservatoire National des Arts et Métiers) a French public institute created in 1794 and dedicated to long-life education. CNAM is based in the heart of Paris and is associated with the museum of arts and crafts. We had 35 submissions and the Program Committee accepted 17 papers. Every submission was assigned to three members of the Program Committee for review. The accepted papers originate from: Algeria, Australia, China, Colombia, France, Germany, India, South Korea, Luxembourg, Morocco, Norway, Tunisia, and United Kingdom. Two brilliant invited speakers completed the technical program. The first speaker was Dr. Hanène Maupas from OT-Morpho, who presented the vision of industry in terms of identity and security in the Internet of Things. The second speaker was Dr. Nikolaos Georgantas, the head of the MIMOVE team at Inria, which is the best research institute in computer science in France. We would like to thank the authors for their high-quality paper contributions, the chairs and the members of the Technical Program Committee (and the additional reviewers) for reviewing the submitted papers and selecting a high-quality program, and the general chairs for their support. Our special thanks go also to the Organizing Committee members for their great help and to the sponsoring institutions. We hope that all the participants enjoyed this successful conference, made a lot of new contacts, and had a pleasant stay in Paris.

CAPE, which stands for Checkpointing-Aided Parallel Execution, is an approach based on checkpoints to automatically translate and execute OpenMP programs on distributed-memory architectures. This approach demonstrates high-performance and completes compatibility with OpenMP on distributed-memory system. This paper presents a new design and implementation model for CAPE that improves the performance and makes CAPE even more flexible.

LTE (Long Term Evolution) was developed and standardized by the 3GPP (3rd Generation Partnership Project). It is a packet switched network. This means that Voice over LTE (VoLTE) is a VoIP service with guaranteed quality of service requirements instead of transmitting in a circuit-switched network such as existing systems (2G/3G). VoLTE is deployed in a fully IP network combined with IMS (IP Multimedia Subsystem). Because of this, the deployment of VoLTE is quite complex and how to ensure the quality of voice transmission over LTE networks is a very big challenge. Thus, several different solutions are needed to enhance and improve the voice transmission quality over LTE networks. In this thesis, we present solutions to improve the voice transmission quality over LTE networks for both narrowband and broadband audio services. For this, we will need different complete factors in solutions. One of them is QoE (Quality of Experience) which is a new trend. And in order to determine the user perception for the real-time service such as VoLTE, we use the extended E-model and WB (wideband) E-model for narrowband and wideband audio services respectively. The solutions proposed here focus on key elements in LTE networks, such as string coding, MAC (Medium Access Control) scheduling systems and monitor voice quality described as follows. First, improved algorithms to enhance the LTE chain codec (encoder and decoder) have been proposed. To enhance the LTE chain coder, a joint adaptation algorithm has been deployed. The goal of this algorithm is to minimize the redundancy generated by chain coding with a slight reduction in the user's perception. Then, in order to improve the LTE chain decoder, an improved Log-MAP algorithm was presented. This algorithm aims to achieve BER (Bit Error Rate) performance that is closest to Log-MAP with reduced computational complexity compared to the state of the art. Secondly, the enhanced user perception and VoIP priority mode scheduling QoS chain and systems have been proposed. These planners are deployed for both wideband and narrowband audio users. Numerical results show that they outperform several featured schedulers such as FLS, M-LWDF and EXP/PF in terms of delay, packet loss rate, cell rate, index and fairness and spectral efficiency in almost all cases. Finally, to ensure voice quality transmission over the LTE network, prediction of user satisfaction is essential. For this reason, we present two non-intrusive models for measuring voice quality on LTE networks. These models are used for both narrowband and broadband audio users. The proposed models do not refer to the original signal. Therefore, they are very suitable for predicting the voice call quality on LTE networks.

The rapid deployment of new infrastructures based on network virtualization and cloud computing triggers new applications and services that in turn generate new constraints such as security and/or mobility. The International Conference on Mobile, Secure and Programmable Networking (MSPN) is aimed at providing a top forum for researchers and practitioners to present and discuss new trends in networking infrastructures, security, services, and applications while focusing on virtualization and cloud computing for networks, network programming, software-defined networks (SDN) and their security. In 2016, MSPN was hosted by CNAM Paris, which is one of the oldest teaching centers in Paris.

No summary available.

This article describes a stateless authentication protocol designed for sensor networks. A mutual authentication between a sensor and a sink can be useful in many applications such as the monitoring of electricity meters or surveillance and monitoring of industrial plants. The authentication protocol we propose can counter usual attacks on sensor networks. First, as it is based on a PUF function, it is efficient against physical node capture. An attacker can not get into the hardware of the sensor or the sink node to obtain the secret keys of the system even if the node is captured physically. Secondly, this protocol can also counter replay attacks since the authentication uses a time-stamp, and the time interval during which a replay attack could be launched can be controlled and greatly reduced. Thirdly, this protocol is stateless, and so the sink can be authenticated to many co-located sensors using a single authentication message. Moreover, the same message can combine the authentication with the transmission of encrypted or unencrypted data.

Software Defined Networking (SDN) has attracted tremendous interest in the telecommunication industry due to its ability to abstract, manage and dynamically re-configure end-to-end networks from a centralized controller. Though SDN is considered to be a suitable candidate for various use cases in mobile networks, none of the work so far has discussed its advantages and actual realization for Train-to-Wayside Communication System (TWC). In this paper, for the first time, the architecture and use cases of SDN controlled mobile backhauling framework for TWC is proposed. We discuss how our proposed architecture can efficiently handle mobility management and also provide dynamic quality-of-service (QoS) for different services on board. As a first step, a software prototype is developed using industrial standard OpenDayLight SDN controller to have our architecture evaluated. Since the automotive sector is being considered to be an important driver for 5G network, our SDN based mobile backhauling solution can be positioned in 5G where SDN plays an important role.

Location-based routings for Vehicular Ad-hoc Networks (VANETs) use location information in routing decisions. However, location-based routing protocols need location services to query the location information of the communication partner node so that packets could be forwarded properly. In this paper, we propose a proactive flooding-based location service which name is Semi-Flooding Location Service (SFLS). The basic design objective in our proposal is to minimize the number of update packets sent over the whole network. We employed a conditional update technique that manages the number of location updates at the forwarder nodes. This significantly reduces the overhead of location updates. To study the effectiveness of our algorithm, a mathematical model is developed, and some numerical results are provided. Results demonstrated that SFLS achieves its design goals.

The goal of cloud federation strategies is to define a mecha- nism for resources sharing among federation collaborators. Those mecha- nisms must be fair to guaranty the common benefits of all the federation members. This paper focuses on intermediate data allocation cost in fed- erated cloud storage. Through a federation mechanism, we propose a mixed integer linear programming model (MILP) to assist multiple data centers hosting intermediate data generated from a scientific community. Under the constraints of the problem, an exact algorithm is proposed to minimize intermediate data allocation cost over the federated data centers storage, taking into account scientific users requirements, inter- mediate data dependency and data size. Experimental results show the cost-efficiency and scalability of the proposed federated cloud storage model.

MPI (Message Passing Interface) and OpenMP are two tools broadly used to develop parallel programs. On the one hand, MPI has the advantage of high performance while being difficult to use. On the other hand, OpenMP is very easy to use but is restricted to shared-memory architectures. CAPE is an approach based on checkpoints to allow the execution of OpenMP programs on distributed-memory architectures. This paper aims at presenting both an in-depth analysis and an evaluation of the performance of CAPE by comparing the execution model of both CAPE and MPI. Some suggestions are also provided to improve the use of CAPE.

No summary available.

How to easily exploit the performance of network using multi-core processors nodes is the purpose of many researches includ- ing CAPE (Checkpointing Aided Parallel Execution). CAPE uses the checkpointing technique to bring the simplicity and high performance of OpenMP - a high performance and easy-to-use standard of parallel programming API on shared-memory architecture - onto distributed- memory architectures. Theoretical analysis and experimental results have proved that CAPE has ability of providing a high performance and complete compatibility with OpenMP standard. This article aims at introducing how to use multiple processes on calculating nodes to increase performance of CAPE with the initial results.

The growth of connected devices, mainly due to the large number of Internet of Things (IoT) deployments and the emergence of mobile cloud services, introduces new challenges for the design of service architectures in the Mobile Cloud Computing (MCC) of mobile cloud computing. In this thesis, we show how SOA can be a key solution for providing distributed mobile cloud services and how the OSGi platform can be an adaptive and efficient framework for providing such an implementation. We adapt the proposed CCM framework to different architectural contexts. The first is a traditional centric model, where mobile devices are reduced to consuming services. The second is a distributed model where the power of mobile-to-mobile interaction provides unlimited opportunities for valuable services, and finally, the three-tier architecture is considered with the introduction of the notion of cloudlet. For each context, we explore the performance of our service-oriented framework and compare it to other existing solutions.

Researchers have been actively working for the past few decades in parallelizing programs so as to cut through massive data chunks for faster response. Current day processors are faster and have more number of cores. So as to utilize the computational capabilities of the processors to its full extend, processes need to be run in parallel. There are several advantages for parallel programs over sequential programs. In sequential programming, the processes execute in a sequential order one after the other. But in parallel programming, we have multiple processes and threads that execute simultaneously at the same time. A task can be performed in lesser time by using parallel programming. But writing a parallel programming manually is a difficult and time consuming task. So we have to use tools to convert a sequential program to a parallel one automatically. OpenMP (Open Multi- Processing) is a set of directives which can be used to generate parallel programs written in C, C++, and FORTRAN to an efficient parallel program. CAPE ( Checkpointing Aided Parallel Execution) is a new paradigm that uses checkpointing technique to generate parallel programs from sequential programs provided with OpenMP directives. is a programming model for performing parallel processing. The main purpose of this paper is to compare the performance and coding complexity of against CAPE under different levels of difficulties.

CAPE - which stands for Checkpointing Aided Parallel Execution - has demonstrated to be a high-performance and compliant OpenMP implementation for distributed memory systems. CAPE is based on the use of checkpoints to automatically distribute jobs of OpenMP parallel constructs to distant machines and to automatically collect the calculated results on these machines to the master machine. However, on the current version, the data exchange on networks use manual sockets that require time to establish connections between machines for each parallel construct. Furthermore, this technique is not really reliable due to the risk of conflicts on ports and the problem of data exchange using stream. This paper aims at presenting the impact of using MPI to improve the reliability and the performance of CAPE. Both socket and MPI implementations are analyzed and discussed, and performance evaluations are provided.

In Cloud storage of multiple CPU cores, many Mapreduce applications may run in parallel on each compute node and collocate with local Disks storage. These Disks storage are shared by multiple applications that use full CPU power of the node. Each application tends to issue contigu- ous I/O requests in parallel to the same Disk. however if large number of Mapreduce tasks enters the I/O phase at the same time, the requests from the same task may be interrupted by the requests of other tasks. Then, the I/O nodes receive these requests as non-contiguous way under I/O con- tention. This interleaved access pattern causes performance degradation for Mapreduce application, this is particularly important when writing intermediate files by multiple tasks in parallel to the shared Disk stor- age. In order to overcome this problem, we have proposed approach for optimizing write access for Mapreduce application. The contributions of this paper are: 1) analyze the open issues on scheduling access request of Mapreduce workload. 2) propose framework for scheduling and predicting I/O request of Mapreduce application. 3) describe each role of compo- nent that intervenes in the scheduling theses I/O request on Block-level of storage server to provide contiguous access.

Mastering the characteristics of the server workload is a key prerequisite in planning and good management of resources within a datacenter. However, for the purpose of research, it is very difficult to find reliable data that can be used to define models adapted to operations in data centers and tests. Google unwounded a portion of this problem by recently publishing a collection of traces from one of its biggest data centers. We conducted a series of analyzes on those data and it basically shows that a large part of the server load relates to non-priority applications and the amount of resources (mainly processors) consumed by servers is much higher than those required for the processing of the received requests. It also highlights the huge amount of wasted energy during processing in those types of data centers.

The rapid deployment of new infrastructures based on network virtualization and cloud computing triggers new applications and services that in turn generate new constraints such as security and/or mobility. The International Conference on Mobile, Secure and Programmable Networking (MSPN) aimed at providing a top forum for researchers and practitioners to present and discuss new trends in networking infrastructures, security, services, and applications while focusing on virtualization and cloud computing for networks, network programming, software-defined networks (SDN) and their security. In 2015, MSPN was hosted by CNAM Paris, which is one of the oldest teaching centers in Paris. The call for papers resulted in a total of 36 submissions from around the world. Every submission was assigned to at least three members of the Program Committee for review. The Program Committee accepted 14 papers, which are from: Algeria, China, Colombia, Denmark, France, Germany, Greece, India, Ireland, Russia, Spain, and Vietnam. One intriguing keynote from Professor Pierre Paradinas complemented the technical program. We would like to thank all who contributed to the success of this conference, in particular the members of the Program Committee (and the additional reviewers) for carefully reviewing the contributions and selecting a high-quality program. Our special thanks go to the members of the Organizing Committee for their great help. We would like to especially thank Habiba Chelah, Lamia Essalhi, and Lynda Saad for taking care of the local arrangements and many other aspects in the organization of the conference. We hope that all participants enjoyed this successful conference, made a lot of new contacts, engaged in fruitful discussions, and had a pleasant stay in Paris, France.

The professional mobile radio communications (PMR) network, which is used for public safety operation, must evolve to broadband solutions to meet the demands of users in the future. In the current broadband technologies, Long Term Evolution (LTE), developed by the 3GPP (3rd Generation Partnership Project), is considered one of the potential candidates for the next generation of PMR. Despite the fact that LTE technology supports high-speed data transmission and supports different packet sizes using adaptive modulation and coding (AMC), LTE is not yet optimized for low-speed voice communication, especially when using LTE for professional mobile radio communications (PMR) context. Therefore, in this thesis, we present solutions to enhance the voice capacity of LTE technology in the PMR context both uplink and downlink transmission. The new proposals, based on the existing LTE standard technology with adaptations, enable the reduction of data overhead and control overhead on LTE (VoLTE) in the PMR context.

This thesis presents several extensions to positive affine interest rate models. A first chapter introduces the concepts related to the models used in the following chapters. It details the definition of so-called affine processes, and the construction of asset price models obtained by non-arbitrage. Chapter 2 proposes a new estimation and filtering method for linear-quadratic state-space models. The next chapter applies this estimation method to the modeling of interbank spreads in the Eurozone, in order to decompose the fluctuations related to default and liquidity risk. Chapter 4 develops a new technique to create multivariate affine processes from their univariate counterparts, without imposing conditional independence between their components. The last chapter applies this method and derives a multivariate affine process in which some components can remain at zero for extended periods. Incorporated into an interest rate model, this process can efficiently account for zero-bottom rates.

No summary available.

No summary available.

This paper presents a Price-Reward learning scheme to encourage mutual coordination between mobile nodes and their wireless networks. In order to maximize the overall network coverage through cooperative diversity, a Nash-Stackelberg Multiplicative Weighted Imitative CODIPAS-RL scheme is proposed based on our previous work. The wireless network implements a 2-level Stackelberg game by introducing Price-Reward (lambda,µ) parameters whereas the Reinforcement Learning (RL) scheme paves the way for mobile nodes to reach a Nash-Equilibrium state. The performance evaluation of the learning scheme for the presented scenario proves fast convergence towards the optimal solution by adopting different sets of actions for the selected strategies. This ensures QoS sustainability during handover situations by data relaying and avoids collisions among mobile nodes while accessing network resources.

Sensor Networks (SN) have been developed for several domains which makes each person or organization has heterogeneous sensors. When considering that the data generated by sensor nodes should be permanently stored and permanently accessible from all over the world by scientists or doctors who need to be notified when some events are triggered, the data sharing, publication and notification become a crucial challenge. This paper present a convergence of social networks, cloud computing and sensor networks to resolve the aforementioned requirements while ensuring privacy and security of users.

Today's Internet has gone through a series of evolutionary changes in the last forty or fifty years. It was designed as a network with fixed nodes. At first, the communication model of the Internet was based on the telephone network (considered 1st Generation Internet). Later it was updated as a client-server model where the communication systems exchange data over dedicated links. This 2nd generation Internet, over the years, has been challenged by many problems such as network congestion, path failure, DOS attacks, mobility management for wireless networks, etc. Internet users are always looking for information, regardless of the location (node or server) where it is located or stored. This approach is the basis of an architecture where information is considered the primary unit. These networks, in general, are called as Network of Information (NetInf), where the information takes a centered position replacing the node-centric approach as in the Internet today. The problems faced by the Internet today, mentioned above, can be addressed with a unifying approach by putting information at the center of the network architecture. On a global scale, this network architecture design is called the Future Information Centric Internet. In parallel, the use of mobile Internet has been increased during the last decade. There were about 1.2 billion mobile broad band subscriptions for 2.4 billion Internet users in 2011. Due to increased spectral efficiency and ubiquitous availability of cellular connectivity, mobility and seamless connectivity is now considered a commodity in everyday life. Nevertheless, in case of the Internet, IP-based mobility solutions cannot catch up in performance with the rapid evolution of cellular networks. Therefore, one of the main goals for the future Internet is to design mobility management systems that overcome the problems in wireless networks such as handover and location management, multihoming, security, etc. In this thesis, we have proposed a solution for mobility management in wireless networks in the context of Information Centric Networking (ICN) in general and in the context of NetInf in particular. NetInf is a Future Internet architecture based on the ICN concept. We propose a mobile node called NetInf Mobile Node (NetInf MN). The architecture of this node is compatible with the TCP/TP based Internet architecture. This architecture design works in conjunction with the Central Control Unit (CCU) to improve handover performance in wireless networks. The Virtual Node Layer (VNL) algorithm explains how the different modules of NetInf MN and CCU units work together. The mathematical model based on Game Theory and Reinforcement Learning (CODIPAS-RL) shows how handover and data relaying are handled in wireless networks. Simulation results show that the proposed model achieves both Nash and Stackelberg equilibria while the CODIPAS-RL scheme reaches a global optimum. Finally, as an example of use case of the NetInf architecture, we propose the NetInf Email Service which does not require dedicated servers and ports unlike the existing email service. The use of asymmetric keys such as user ID is the unique feature proposed for this service. The NetInf Email service architecture presented, explains how different architectural elements work together. We discuss the different challenges and requirements for this service. The prototype developed for NetInf will be used for the implementation of this service.

The development of wireless sensor networks means that each user or organization is already connected to a large number of nodes. These nodes generate a large amount of data, making the management of this data non-obvious. Moreover, this data may contain privacy information. The work in this thesis addresses these issues. First, we designed a middleware that communicates with physical sensors to collect, store, translate, index, analyze and generate alerts on sensor data. This middleware is based on the notion of components and composites. Each physical node communicates with a composite of the middleware via a RESTFul interface. This middleware has been tested and used in the framework of the European project Mobesens in order to manage the data of a sensor network for water quality monitoring. Secondly, we designed a hybrid pair and group authentication and key establishment protocol. Considering that there is a performance difference between the sensor nodes, the gateway, and the middleware, we used identity-based cryptography-based authentication between the gateway and the storage server and symmetric cryptography between the sensors and the other two parties. Then, the middleware was generalized in the third part of the thesis so that each organization or individual can have their own space to manage their sensor data using cloud computing. Then, we have secure social portal for sharing sensor network data.

In order to exploit the capabilities of parallel architectures such as clusters, grids, multi-processor systems, and more recently clouds and multi-core systems, a universal and easy-to-use programming language remains to be developed. From a programmer's point of view, OpenMP is very easy to use largely because of its ability to support incremental parallelization, the ability to dynamically define the number of execution threads, and also because of its scheduling strategies. However, because it was originally designed for shared memory systems, OpenMP is generally very limited for performing computations on distributed memory systems. Many solutions have been tried to run OpenMP on distributed memory systems. The most successful approaches focus on exploiting a special network architecture and therefore cannot provide an open solution. Others are based on an already available software solution such as DMS, MPI or Global Array, and therefore have difficulties in providing a fully compliant, high performance implementation of OpenMP. CAPE - for Checkpointing Aided Parallel Execution - is an alternative solution for developing a compliant OpenMP implementation for distributed memory systems. The idea is the following: when arriving at a parallel section, the image of the master thread is saved and sent to the slaves. Then, each slave executes one of the threads. At the end of the parallel section, each slave thread extracts a list of any modifications that have been made locally and sends it back to the master thread. In order to prove the feasibility of this approach, the first version of CAPE was implemented using full recovery points. However, a preliminary analysis showed that the large amount of data transmitted between threads and the extraction of the list of changes from full resume points leads to poor performance. Moreover, this version is limited to parallel problems satisfying Bernstein's conditions, i.e., it does not allow for shared data. The objective of this thesis is to propose new approaches to improve the performance of CAPE and overcome the restrictions on shared data. First, we developed DICKPT (Discontinuous Incremental ChecKPoinTing), an incremental checkpointing technique that supports the ability to take discontinuous checkpoints during the execution of a process. Based on DICKPT, the execution speed of the new version of CAPE has been increased considerably. For example, the time to calculate a large matrix-matrix multiplication on a cluster of desktops has become very similar to the runtime of an optimized MPI program. Furthermore, the speedup associated with this new version for various numbers of threads is quite linear for different problem sizes. For shared data, we proposed UHLRC (Updated Home-based Lazy Relaxed Consistency), a modified version of the HLRC (Home-based Lazy Relaxed Consistency) memory model, to make it more suitable for CAPE characteristics. Prototypes and algorithms to implement data synchronization and shared data directives and clauses are also specified. These two works guarantee the possibility for CAPE to respect OpenMP shared data requests.

Developed since the beginning of the 90s, PC clusters are increasingly becoming an alternative to supercomputers. In particular, the recent advent of gigabit networks has reinforced this position. It is in this context that the multi-PC machine was developed, implementing the remote write protocol (where the local and remote physical addresses must be specified by the message sender). If this protocol is very efficient, the use of physical addresses by the user constitutes a very important security hole. The organization of contiguous physical memory blocks͏̈, The protection of these blocks through a signature and/or an encryption, original efficient mechanisms adapted to this information, the fine evaluation of the various times intervening in the transmissions are the main contributions of this thesis.

Development associates with differentiation, a control of cell proliferation that takes place throughout the life of the organism. Morphogenesis reflects the existence of interactions between these two levels of cell physiology that are also found associated in pathological situations (malignancy). This work is a study of the effect of a transient inhibition of proliferation on the phenotypic properties of variant cells of a rat hepatoma line in in vitro culture. The original h4iiec3 line cells express many hepatocyte functions and induce tumors in compatible animals. The c2 variant clone is characterized by loss of the differentiated state and tumorigenicity. Partial copper deficiency applied to c2 cells results in: (1) transient inhibition of cell proliferation. (3) induction of reporter gene amplification events, consistent with a dissociation between cell proliferation and dna synthesis observed under these conditions. (4) cell death by apoptosis recorded at the time of reversion induction. A second series of experiments exploits the association of tumorigenicity with the ability of cells to grow in vitro in semi-solid medium. Under these culture conditions, a fraction of the c2 cells develop colonies. Once acquired, the ability to multiply in semi-solid medium is stable. These c2ag cells are characterized by an increased reversion frequency and by the expression of transcription factors enriched in the liver. All the data acquired reveal the inducing effect of a transient inhibition of proliferation on the generation of stable differential revertants. These results also suggest that a key step in the transition to the malignant state involves the involvement of transformed cells in the differentiation process. The nature of the observations made on this system reveals the non-linear character of the mechanisms involved in the dynamics of cellular responses and thus the extreme sensitivity to initial conditions.

The activity of the antitumor drug pazelliptin (pze) does not seem to be due only to its intercalation in dna, but would require a metabolic activation leading to the formation of radical species. A part of this work deals with the study of the reaction dynamics of such species in solution and towards dna. Experimentally, pze radicals are generated by pulsed radiolysis or laser-photolysis methods using the possibility of two-photon photoionization of the drug. These two complementary studies have allowed us to characterize by transient absorption spectroscopy measurements, the reactivity of the pze radicals in these different protonation states and to propose a mechanism of action of the drug towards dna involving its radical species and the active oxygen species. Furthermore, we have shown that pze is a photosensitizer of dna degradation by measuring single and double strand breaks. The investigation of the molecular mechanisms responsible for this process led us to characterize the photophysical and photochemical properties of the triplet state of pze in solution and complexed to nucleic acids. The results allowed us to propose a mode of action of the drug through a charge transfer oxidation of dna from the triplet state of pze.

No summary available.

No summary available.