Archive ouverte de l'ENAC
Not a member yet
3458 research outputs found
Sort by
Causette: User-Controlled Rearrangement of Causal Constructs in a Code Editor
No full textInternational audienceProgramming interaction usually involves specifying causal relationships such as input events triggering a state change or the propagation of values. Such code may reside in several locations and its execution is driven by multiple causal chains, which hinders the programmer’s ability to understand and fix it. We designed Causette, a set of four novel interaction techniques for a code editor. They consist in rearranging causal constructs on demand to makethe code representation consistent with the causal chain being analyzed by the user.We ran an experiment showing that Causette may be more usable than a regular editor for some code understanding tasks. This work suggests that rearranging interaction code may help developers better understand and fix it
Sécurisation des communications dans un réseau ad hoc au sein d’un essaim de drones
No full textDrones become more and more frequent in our everyday life as a leasure and also in the industry. Analysts forcast a steady growth of the civilian drones market which could reach 10 to 20 billion euros worldwide in the coming years.Nowadays, missions mostly operate single Unmaned Aerial Vehicle (UAV). But researchers are now considering swarms of drones to be more efficient to solve specific problems. Drones now have to collaborate and coordinate.Teams of drones require the availability of a wireless network for all the tasks required by the mission but also for additional coordination and synchronisation needs.Wireless networks are open to the outside by nature and securing such network is a challenging task. Several solutions proposed to tackle this issue from different aspects of data communication and securing either the physical layer or the routing protocols, or at the application level in multi-agent systems. None, however, considered securing the access to the network and few proposed efficient counter-measures.In this thesis, we propose a security oriented network architecture that allows controlling the communication in the network, with no fixed ground based infrastructure and with a single wireless interface card. It brings emerging Software Defined Network technology together with AODV, a routing protocol suitable for flying ad hoc network (FANET). We demonstrate that the architecture allows to protect the network against most sorts of attacks from the outside. In addition, it brings a good knowledge of the activity within the network, which is a prerequisite to further improve security.From this knowledge, we propose a detection algorithm for traffic injection attacks from an outside node and the corresponding counter-measure. Then, we propose a set of measurable features on the activity within the network suitable for a machine learning algorithms to detect abnormal behaviors.We demonstrate the relevance of these features by training a Random Forest classification machine learning algorithm on a dataset consisting of network captures including several network attacks : denial of service (DoS), port scan, password cracking using bruteforce and distributed denial of service (DDoS). The performances of the detection based on these features are promissing, not only in terms of precision but also in terms of speed, paving the way for applying counter measures in real time. The latter may be conveniently put in place using the proposed architecture. Tests using representative scenarios of network traffic for a swarm of drone show a good generalization of the ML model and good performances.Les drones sont de plus en plus présents, dans nos vies pour le loisir comme dans l'industrie. Les prévisions sur le marché des drones civils envisagent une croissance importante sur les prochaines années et pourrait atteindre 10 à 20 milliards d'euros au niveau mondial.Si les missions confiées aux drones ont tout d'abord considéré des drones isolés, certains types de missions nécessitent la collaboration de plusieurs d'entre eux au sein d'une flotte.Une flotte de drones nécessite la mise en œuvre et la disponibilité d’un réseau sans fil pour toute les tâches ayant trait d’une part à la mission et d’autre part à toute coordination ou synchronisation. Les réseaux sans fil sont par nature ouverts sur l’extérieur et il se pose donc la question de leur sécurisation. Plusieurs travaux de recherche ont abordé cette question avec différents angles d’attaque : la couche physique, les protocoles de routage, les systèmes multi agents. Mais aucun n’aborde la question de la sécurisation de l’accès à ce réseau et peu ont étudié la question des réponses à apporter en cas d’attaque.Dans cette thèse nous proposons une architecture orientée vers la sécurité permettant une meilleure maîtrise des communications dans le réseau, et s'affranchissant entièrement de toute infrastructure fixe au sol. Cette architecture allie les réseaux définis par logiciels (SDN), qui est une technologie qui a émergé récemment, avec AODV, un protocole de routage adapté aux réseaux ad hoc de type FANET. Nous démontrons que cette architecture permet de protéger le réseau contre la plupart des attaques depuis l'extérieur. Cette architecture nous permet également d'obtenir une bonne connaissance de l'activité dans le réseau, pré-requis pour améliorer la sécurité.De cette connaissance, nous proposons d'une part une technique de détection d'injection de trafic depuis l'extérieur et une méthode pour s'en défendre. D'autre part, nous proposons un ensemble de caractéristiques mesurables de l’activité du réseau propres à être utilisées avec un algorithme d’apprentissage automatique.Nous démontrons la pertinence de ces mesures en entraînant un modèle de classification par apprentissage supervisé de type Random Forest sur un ensemble de captures réseaux présentant des attaques sur le réseau: déni de service (DoS), balayage de ports, découverte de mot de passe (brute force) et déni de service distribué (DDoS). Les performances en terme de détection d’attaques basées sur ces caractéristiques sont prometteuses, non seulement en terme de précision mais également en terme de vitesse de détection, offrant ainsi la possibilité d'une réaction en temps réel. Cette réaction peut être mise en œuvre grâce à l'architecture proposée dans cette thèse. Des tests sur des scénarios représentatifs d'un trafic réseau pour une flotte de drones montrent que le modèle est capable de généraliser avec de bonnes performances sur notre cas d'étude
Quelques raisons montrant que l'ordinateur quantique ne sera pas utile
No full textInternational audienc
A heuristic design grid for past and future uses of Token+Constraint systems
No full textInternational audienc
Optimising security screening resources during airport access mode disruptions
No full textInternational audienceAirport access mode disruptions have a significant impact on passenger arrival times and thus on congestion level at security screening systems. During such events, information sharing between ground and air transportation stakeholders could be a key lever to optimize airport operations at a tactical level. In this work, an online reallocation of airport security teams across the different security screening checkpoints is considered. Three integer linear programming formulations of the problem are proposed to increase the level of service of an airport when a disruption occurs. A study case based on one day of operations at Paris-Charles de Gaulle airport is considered. Results show that reallocating airport security staff when outbound passengers are delayed could significantly improve airport security system performances. The different allocations obtained lead to a drop in the maximum waiting time up to 72%. In addition, the average waiting time and the number of stranded passengers at the airport are reduced
Simulated-Annealing Hyper-Heuristic for Demand-Capacity Balancing in Air Traffic Flow Management
No full textInternational audienceVery critical convective weather leads to sharp drop in air-traffic en-route sector capacity keen on creating severe demand-capacity imbalances, called also hotspots. Following the knock-on effect, these imbalances spread across the network, causing the so-called netspots. The problem of demand-capacity hotspot/netspot mitigation with demand-side measures aims at solving these imbalances using ground delays and reroutings, while minimizing the total delay at arrival. A simulated-annealing hyper-heuristic relying on efficient neighborhood operators is proposed. A study on the combination of different delay-based and rerouting-based neighborhood operators is conducted. The best combination is used to solve a large-scale and challenging instance, in a relatively short computation time
Evaluating Aviation Emission Inefficiencies and Reduction Challenges with Electric Flights Based on an analysis of flights from 2019 in the Dutch and French airspaces
No full textInternational audienceInefficiencies in flight operations, like deviations and non-optimal flight speed or altitude, are directly linked to flight emission inefficiencies. Quantifying these emission inefficiencies and studying potential mitigation strategies is certainly beneficial for the sustainability of the aviation industry. In this paper, we analyze emission inefficiencies in Dutch and French airspaces using flight data from 2019. The emission inefficiency analysis quantifies the excess carbon emissions for each flight by comparing its emissions with a set of optimal alternative trajectories. We find that around 19% of excess emissions existed in 2019 within the airspace of interest. We also study the potential reduction of emissions by replacing short-range flights with electric aircraft. We propose a simple electric aircraft energy model and relate that to emissions in electric generations in different countries. We find that besides the significant increase in air traffic demand caused by the limited capacities of electric flights, the emissions caused by electricity generation cannot be neglected. Significant reductions can only be achieved when emissions caused by electricity generation are low, as is the case currently in France. However, more emissions can be indirectly generated if the electricity used to power the future electric aircraft is itself produced from high emission sources, as is the case currently in the Netherlands. The paper also provides further insights and recommendations on the data sources, research approach, and future research for aviation sustainability
Exploiting spatio-temporal partial separability of large-scale airspaces
No full textInternational audienceThis paper addresses large-scale flight planning via a divide-and-conquer technique that exploits the partial separability feature of the problem. 4D-interaction between flights is used to cluster the flights, and these clusters are then exploited to improve the optimization process. Preliminary computational experiments on the French airspace demonstrate the natural separability of air traffic and yield promising computational improvement for flight planning thanks to the clustering
Uncertainty Inclusive Runway Balancing Using Convolutional Neural Network
No full textInternational audienceThis paper proposes a new optimization scheme using neural networks for runway balancing to minimize departure and arrival aircraft delay. The delay prediction for runway balancing optimization is obtained by a neural network, only without any additional simulations. Developing an accurate simulation model under an uncertain environment is difficult, but the proposed neural network model can estimate the average delay without modeling uncertainty explicitly. In this paper, the effectiveness of the proposed method is validated through numerical simulations. First, simulations are used to generate the data, which are then used to train the neural network. Next, the runway balancing problem is solved via simulated annealing using the delay predicted by the neural network. The simulation result shows that the proposed approach outperforms the simulation-based method under an uncertainty environment. Therefore, the neural network is shown to accurately estimate the delay under the uncertainty environment, which makes the proposed neural-network-based method applicable to objective function calculations for optimization
Re-entry predictions of space debris for collision avoidance with air traffic
No full textInternational audienceAbstract The need for a safe and efficient integration of space vehicle operations into air traffic system to minimize the risk of impacts of spacecraft and aircraft and to sustain a steady air traffic is evident. This work provides a strategy toward more efficient management of uncontrolled re-entries by combining uncertainty propagation analysis with the FAA’s ConOps in the framework of the FAA’s Next Generation Air Transportation System (NextGen). The paper considers the scenario where a spacecraft re-entry is completely uncontrolled. During such re-entry, predictions are very difficult and are affected by various sources of uncertainty. Then the resulting position distribution throughout airspace boundaries is analysed and the impact on air traffic is estimated by defining protected airspace areas. The impact of the size of the protected area on the air traffic control is analysed and strategies for re-routing of air traffic are proposed. To verify the proposed results, the re-entry of the core stage of the CZ-5B R/B launcher is simulated