Since the scheduling algorithm is not standardized for IEEE 802.15.4e TSCH networks, many scheduling algorithms have been proposed. Most of them are centralized, few are distributed. Among the distributed scheduling algorithms, many rely on assumptions that may be violated by real deployments. This violation usually leads to conflicting transmissions of application data, decreasing the reliability and increasing the latency of data delivery. Others require a processing complexity that cannot be provided by sensor nodes of limited capabilities. Still others are unable to adapt quickly to traffic or topology changes, or are valid only for small traffic loads.

In the study funded by the Inria ADT DASMU (Action de Developement Technologique Distributed Adaptive Scheduling for MUltichannel wireless sensor networks), we focus on a distributed scheduling algorithm that relies on realistic assumptions, does not require complex computation, is valid for any traffic load, is adaptive and compliant with the standardized protocols used in the 6TiSCH working group at IETF.

First results have been obtained and an intensive simulation campaign made with the 6TiSCH simulator has provided comparative performance results. Our proposal outperforms MSF, the 6TiSCH Minimal Scheduling Function, in terms of end-to-end latency and end-to-end packet delivery ratio. More evaluations are needed to improve the proposal (e.g. less packet drops during transient situations, less overhead) in terms of scheduled cells).

In 2018, we focus on how an IEEE 802.15.4e is autonomously built and how nodes join the network.

To join the TSCH network, a device randomly selects a physical channel used by this network and listens to a beacon advertising this network. Since the physical channel on which the beacon is broadcast changes at each beacon slot due to channel hopping, the joining device will eventually hear a beacon sent by one of its neighbors. Upon receipt of a valid beacon, this device gets synchronized with the TSCH network.

In this study, we focus on the time needed by a node to detect a beacon sent by a TSCH network, as well as on the time needed to build a TSCH network. These times are important for industrial applications where new nodes are inserted progressively, or when failed nodes are replaced. Both times highly depend on the beacon advertisement policy, policy that is not specified in the standard and is under the responsibility of a layer upper than the MAC one. Since beacons are broadcast, they are lost in case of collisions: the vital information they carry is lost. The main problem is how to avoid collisions between two devices that are not neighbors.

That is why we propose the Enhanced Deterministic Beacon Advertising algorithm, called EDBA, that ensures a collision-free advertising of beacons. Since the beacon cells are fairly distributed in the slotframe, the average joining time is minimized. The behavior of a joining node has been modeled by a Markov chain from which the average joining time is computed, taking into account the reliability of wireless links. An intensive performance evaluation based on NS3 simulations allows us to validate this model and conclude on the very good performance of EDBA, even when compared with MBS, considered as the best advertising algorithm in the literature. These results have been published in the Annals of Telecommunications, .

The recent advances in wireless sensors and Unmanned Aerial Vehicles have created new opportunities for environmental control and low cost aerial data gathering. We propose to use an Unmanned Aerial Vehicle (UAV) for data gathering . Basically, we have proposed a method for UAV path planning based on virtual forces and potential fields. In addition, and more importantly, we present a new approach to compute the attractive forces of the potential field.

We use as our starting point the idea used by Pereira of using a potential field approach. However, we extend this work by considering that each cell in the area apply an attractive force on the drone, not only the deployed sensors. We compared our results with those obtained with Pereira's method and we obtained better performance in terms of data collection time. In other words, for the same period of time our method collect more data. The second advantage of our approach is that it leads to a significant reduction in the distance that the drone must travel.

Named Data Networking is a promising architecture for emerging Internet applications such as the Internet of Things (IoT). Many studies have already investigated how NDN can be an alternative for IP in future IoT deployments. However, NDN-IoT propositions need accurate evaluation at network level and system level as well. We introduce an NDN framework for OMNeT++ . Designed for low-end devices and gateways of the IoT, the framework is capable of simulating NDN scenarios at the boundary of the network and the system. The framework implementation is presented and used to study a typical aspect of NDN integration in IoT devices.

We present a simple, stochastic-geometric model of a wireless access network exploiting the LoRA (Long Range) protocol, which is a non-expensive technology allowing for long-range, single-hop connectivity for the Internet of Things. We assume a space-time Poisson model of packets transmitted by LoRA nodes to a fixed base station. Following previous studies of the impact of interference, we assume that a given packet is successfully received when no interfering packet arrives with similar power before the given packet payload phase, see . This is as a consequence of LoRa using different transmission rates for different link budgets (transmissions with smaller received powers use larger spreading factors) and LoRa intra-technology interference treatment. Using our model, we study the scaling of the packet reception probabilities per link budget as a function of the spatial density of nodes and their rate of transmissions. We consider both the parameter values recommended by the LoRa provider, as well as proposing LoRa tuning to improve the equality of performance for all link budgets. We also consider spatially non-homogeneous distributions of LoRa nodes. We show also how a fair comparison to non-slotted Aloha can be made within the same framework.

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 Internet of Things (IoT) connects tiny electronic devices able to measure a physical value (temperature, humidity, etc.) and/or to actuate on the physical world (pump, valve, etc). Due to their cost and ease of deployment, battery-powered wireless IoT networks are rapidly being adopted.

The promise of wireless communication is to offer wire-like connectivity. Major improvements have been made in that sense, but many challenges remain as industrial application have strong operational requirements. This section of the IoT application is called Industrial IoT (IIoT).

The main IIoT requirement is reliability. Every bit of information that is transmitted in the network must not be lost. Current off-the-shelf solutions offer over 99.999% reliability.

Then come latency and energy-efficiency requirements. As devices are battery-powered, they need to consume as little as possible to be able to operate during years. The next step for the IoT is to target time-critical applications.

Industrial IoT technologies are now adopted by companies over the world, and are now a proven solution. Yet, challenges remain and some of the limits of the technologies are still not fully understood. In his PhD Thesis, Keoma Brun-Laguna addresses TSCH-based Wireless Sensor Networks and studies their latency and lifetime limits under real-world conditions.

We gathered 3M network statistics 32M sensor measurements on 11 datasets with a total of 170,037 mote hours in real-world and testbeds deployments. We assembled what we believed to be the largest dataset available to the networking community.

Based on those datasets and on insights we learned from deploying networks in real-world conditions, we study the limits and trade-offs of TSCH-based Wireless Sensor Networks. We provide methods and tools to estimate the network performances of such networks in various scenarios. We highlight the trade-off between short latency and long network lifetime. We believe we assembled the right tools for protocol designer to build deterministic networking to the Industrial IoT.

By the year 2020, it is expected that the number of connected objects will exceed several billions devices. These objects will be present in everyday life for a smarter home and city as well as in future smart factories that will revolutionize the industry organization. This is actually the expected fourth industrial revolution, more known as Industry 4.0. In which, the Internet of Things (IoT) is considered as a key enabler for this major transformation. IoT will allow more intelligent monitoring and self-organizing capabilities than traditional factories. As a consequence, the production process will be more efficient and flexible with products of higher quality.

To produce better quality products and improve monitoring in Industry 4.0, strong requirements in terms of latency, robustness and power autonomy have to be met by the networks supporting the Industry 4.0 applications. The wireless TSCH (Time Slotted Channel Hopping) network specified in the e amendment of the IEEE 802.15.4 standard has many appealing properties. Its schedule of multichannel slotted data transmissions ensures the absence of collisions. Because there is no retransmission due to collisions, communication is faster. Since the devices save energy each time they do not take part in a transmission, the power autonomy of nodes is prolonged. Furthermore, channel hopping enables to mitigate multipath fading and interferences.

To increase the flexibility and the self-organizing capacities required by Industry 4.0, the networks have to be able to adapt to changes. These changes may concern the application itself, the network topology by adding or removing devices, the traffic generated by increasing or decreasing the device sampling frequency, for instance. That is why the flexibility of the schedule ruling all network communications is needed.

In 2018, we show how a TSCH network can adapt to such changes. More precisely, we propose a solution ranging from network construction to data gathering. We show how a TSCH network is autonomously built, supports data gathering and is able to adapt to changes in network topology, traffic and application requirements.

The solution proposed preserves the merits of TSCH network, that can be listed hereafter. The time-slotted multichannel medium access enables parallel transmissions on several channels, leading to shorter latency and higher throughputs. In addition, channel hopping mitigates interference and multipath effects. Furthermore, since transmissions are scheduled, a conflict-free schedule is computed by the network coordinator (i.e. the CPAN). Hence, no collision occurs during data gathering. The absence of collision leads to a higher throughput, because there is no retransmission due to collisions. It also preserves nodes power autonomy.

This simple solution is based on the coexistence of several periodic slotframes. We distinguish three slotframes, which are the Beacon Slotframe, the Data Slotframe and the Shared Slotframe. The network schedule corresponds to the superposition of the three schedules given by each slotframe, where the slotframe with the highest priority wins.

This solution ensures a collision-free dissemination over the whole network. Beacons are broadcast in sequence by increasing depth of devices. This broadcast is also used to disseminate Data Schedules (new schedule or update).

In addition, this solution is adaptive. Topology, traffic or application changes are notified to the CPAN. Depending on the changes notified, the CPAN updates the current schedule or recomputes a new one. Shared slots are used to cope with unexpected events.

We compute the theoretical bounds with regard to key performance indicators and compare them with the values obtained by NS3 simulation. Simulation results confirm the theoretical upper bounds computed for network construction and data gathering. Hence, TSCH networks are able to adapt to traffic or topology changes in a reasonable time which is a strong requirement of Industry 4.0 applications. These results have been presented at the PEMWN 2018 conference in . In some further work, we will study how to improve this delay to support the most demanding applications.

In High Performance Computing (HPC), it is assumed that all machines are homogeneous in terms of CPU and memory capacities, and that the tasks making up the jobs have similar resource requests. It has been shown that this homogeneity relating both to machine capacity and workload, although generally valid for HPC, does no longer apply to data centers. This explains why the publication of data gathered in an operational Google data center over 29 days has aroused great interest among researchers.

It is crucial to have real traces of a Google data center publicly available that are representative of the functioning of real data centers. Our goal is to analyze the data collected and to draw useful conclusions about machines, jobs and tasks as well as resource usage. Our main results have been published in  , and can be summarized as follows:

Such results are needed to validate or invalidate some simplifying assumptions that are usually made when reasoning about models, and make the models more accurate for jobs and tasks as well as for available machines. Having validated these models on real data centers, they can then be used for extensive evaluation of placement and scheduling algorithms and more generally for resource allocation (i.e. CPU and memory). These algorithms can then be applied in real data centers.

Another possible use of this data set is to consider it as a learning set in order to predict some feature of the data center, such as the workload of hosts or the next arrival of jobs.

To limit global warming, all industrial sectors must make effort to reduce their carbon footprint. Information and Communication Technologies (ICTs) alone generate 2% of global CO2 emissions every year. Due to the rapid growth in Internet services, data centers have the largest carbon footprint of all ICTs. According to ARCEP (the French telecommunications regulator), Internet data traffic multiplied by 4.5 between 2011 and 2016. In order to support such a growth and maintain this traffic, data centers'energy consumption needs to be optimized. The problem of managing Data Centers (DC) and clouds optimally, in the sense that the demand is met with a minimal energy cost, remains a major issue. In this research, we evaluate the maximum energy saving that can be obtained in DCs by means of a proactive management of resources. The proposed management is based on models that predict resource requests.

Diverse approaches to obtain predictive models of DCs have been studied recently. Among the most popular methods with the comparatively lowest prediction errors are the predictive models of the ARMAX family. Hence, we study the predictive model given by the ARMAX family. We compare its performance with that of the Last Value (LV) model which predicts that the next value will be equal to the current one. To the best of our knowledge, there are no studies relating to the performance bounds that can be achieved using these models. In this research, we study the limits of the improvement in terms of energy cost that can be obtained using proactive strategies for DC management based on predictive models.

Using the Google dataset collected over a period of 29 days and made publicly available, we evaluate the largest benefit that can be obtained with those two predictors.

We study positioning systems using Vehicular Ad Hoc Networks (VANETs) to predict the position of vehicles  . We use the reception power of the packets received by the Road Side Units (RSUs) and sent by the vehicles on the roads. In fact, the reception power is strongly influenced by the distance between a vehicle and a RSU. To predict the position of vehicles in this context, we adopt the machine learning methodology. As a pre-requisite, the vehicles know their positions and the vehicles send their positions in the packets. The positioning system can thus perform a training sequence and build a model. The system is then able to handle a prediction request. In this request, a vehicle without external positioning will request its position from the neighboring RSUs. The RSUs which receive this request message from the vehicle will know the power at which the message was received and will study the positioning request using the training set. In this study, we use and compare three widely recognized techniques : K Nearest Neighbors (KNN), Support Vector Machine (SVM) and Random Forest. We study these techniques in various configurations and discuss their respective advantages and drawbacks. Our results show that these three techniques provide very good results in terms of position predictions when the error on the transmission power is small.

We study the use of the Support Vector Machine technique to estimate the probability of the reception of a given transmission in a Vehicular Ad hoc NETwork (VANET). The transmission takes place between a vehicle and a RoadSide Unit (RSU) at a given distance and with a given transmission rate. The RSU computes the statistics of the receptions and is able to compute the percentage of successful transmissions versus the distance between the vehicle and the RSU and the transmission rate. Starting from this statistic, a Support Vector Machine (SVM) scheme can produce a model. Then, given a transmission rate and a distance between the vehicle and the RSU, the SVM technique can estimate the probability of a succcessful reception. This probability can be used to build an adaptive technique which optimizes the expected throughput between the vehicle and the RSU. Instead of using transmission values of a real experiment, we use the results of an analytical model of CSMA that is customized for 1D VANETs. The model we adopt to perform this task uses a Matern selection process to mimic the transmission in a CSMA IEEE 802.11p VANET. With this model we obtain a closed formula for the probability of successful transmissions. Thus with these results we can train an SVM model and predict other values for other couples : distance, transmission rate. The numerical results we obtain show that SVM seems very suitable to predict the reception probability in a VANET.

We focus on vehicular safety applications based on the Dedicated Short Range Communication (DSRC) standard. We propose a new mechanism to alleviate channel congestion by reducing the beacons load while maintaining an accurate awareness level. Our scheme is based on the collective perception concept which consists in sharing perceived status information collected by vehicles equipped with different types of sensors (radars, lidars, cameras, etc.). To achieve our goal, we propose two main schemes . The first one consists in implementing the collective perception capability on vehicles and adding a new category of status messages to share locally collected sensor data in order to reduce channels load and enhance vehicles' awareness. The second scheme concerns the accuracy level of the received information from the collective perception enabled vehicles by fixing a prior error threshold on the position. The method proposed is validated by simulations and the results obtained are compared to those of an application based on the traditional beaconing scheme of the IEEE802.11p standard. The simulations show that the proposed scheme is able to significantly reduce the load on the control channel incurred by the beacons and the packet error ratio for different network densities and built-in sensors characteristics.

We focus on vehicular safety applications based on the Dedicated Short Range Communication (DSRC) standard. We propose a new mechanism to alleviate channel congestion by reducing the beacons load while maintaining an accurate awareness level. Our scheme is based on the collective perception concept which consists in sharing perceived status information collected by vehicles equipped with different types of sensors (radars, lidars, cameras, etc.). To achieve our goal, we propose two main schemes . The first one consists in implementing the collective perception capability on vehicles and adding a new category of status messages to share locally collected sensor data in order to reduce channels load and enhance vehicles' awareness. The second scheme concerns the accuracy level of the received information from the collective perception enabled vehicles by fixing a prior error threshold on the position. The method proposed is validated by simulations and the results obtained are compared to those of an application based on the traditional beaconing scheme of the IEEE802.11p standard. The simulations show that the proposed scheme is able to significantly reduce the load on the control channel incurred by the beacons and the packet error ratio for different network densities and built-in sensors characteristics.