Institut de l'Information Scientifique et Technique
Portail HAL-LARA (Libre accès aux rapports scientifiques et techniques)Not a member yet
78443 research outputs found
Sort by
Une parcelle du faubourg Saint-Jacques : du clos des Jacobins à la rue Malebranche (XIIIe-XXIe siècle)
LIFT: Byzantine Resilient Hub-Sampling
Recently, a novel peer sampling protocol, Elevator,was introduced to construct network topologies tailored foremerging decentralized applications such as federated learningand blockchain. Elevator builds hub-based topologies in a fullydecentralized manner, randomly selecting hubs among partic-ipating nodes. These hubs, acting as central nodes connectedto the entire network, can be leveraged to accelerate messagedissemination. Simulation results have shown that Elevator con-verges rapidly (within 3–4 cycles) and exhibits robustness againstcrash failures and churn. However, its resilience to Byzantineadversaries has not been investigated.In this work, we provide the first evaluation of Elevator underByzantine adversaries and show that even a small fraction (2%)of Byzantine nodes is sufficient to subvert the network. As aresult, we introduce LIFT, a new protocol that extends Elevatorby employing a cryptographically secure pseudo-random numbergenerator (PRNG) for hub selection, thereby mitigating Byzan-tine manipulation. In contrast, LIFT withstands adversarialinfiltration and remains robust with up to 10% Byzantine nodes.These results highlight the necessity of secure randomness indecentralized hub formation and position LIFT as a more reliablebuilding block for Byzantine-resilient decentralized systems
RESEARCH ON THE INTEGRATION OF INDEPENDENT WASTE WORKERS IN SOLID WASTE MANAGEMENT IN VIETNAM STUDY REPORT
Analytical Comparison Between IEEE 802.11p and IEEE 802.11bd Based on an Extended BianchiModel
IEEE 802.11p has long been the baseline for vehicular communication (V2X) applications, while IEEE 802.11bd has emerged as its successor, promising higher reliability, throughput, and backward compatibility. In this paper, we compare the performance of both standards using an analytical model inspired by Bianchi's saturation throughput analysis. We propose a modification to the original Bianchi model to account for physical layer enhancements in IEEE 802.11bd by introducing a success probability even in the presence of collisions. Specifically, we incorporate a tunable parameter to model the impact of advanced features such as MIMO, higher-order modulation, and cooperative retransmission, which improve the likelihood of packet delivery despite contention. The modified model enables a more accurate analytical comparison between the two standards under saturated traffic conditions. Our analysis reveals the extent to which IEEE 802.11bd mitigates the negative impact of collisions and enhances throughput performance compared to its predecessor
L'agilité organisationnelle dans les réseaux de franchise
Cette recherche commanditée par la Fédération Française de la Franchise a pour ambition de caractériser les pratiques d’agilité organisationnelle développées par les réseaux de franchise pour faire face aux changements de leurs environnements. L’étude s’intéresse également aux adaptations du business model réalisées en réponse aux changements de l’environnement. Durée du contrat : 18 mois Membres de l’équipe : Magali MALHERBE, NIMEC, Université de Caen Normandie, coordinatriceAnne-Laure LE NADANT, LiRIS, Université Rennes 2, coordinatriceAlexandra Burlaud, DRM M-Lab, Université Paris Dauphine PSLFrédéric Perdreaux, COACTIS, Université de Saint-Etienne Sanoussy SOW, METIS, EM Normandi
Quelques problèmes de graphes avec jetons
In graph theory, any vertex of a graph has a given set of incident edges and neighbors. All these objects (vertices and edges) are available and can be used to produce a solution to any graph problem. In this note we explore another model: a vertex only has a limited number of its incident edges or neighbors that can be used. This restricted capability is modeled by tokens given to each vertex that must be assigned to them.In this note we propose the reformulation of several classical problems (vertex-cover, domination, independent domination, connectivity, distance and spanning tree) in this model ; then we propose solutions to solve them or we prove hardness results
Developing historical gridded climate datasets for Viet Nam using reanalysis data and station observations since the 1940s
Accurate and high-resolution gridded climate data are crucial for scientific applications, yet such data have long faced limitations due to sparse and uneven station networks. In this study, we address this gap for Vietnam by developing the Vietnam Historical Gridded Climate (VnHGC) dataset, which for the first time combines an expanded suite of historical station observations, including newly digitized and quality-controlled records with state-of-the-art reanalysis products through Optimal Interpolation (OI). Two products were constructed: a monthly dataset for 1940-1953 with a horizontal resolution of 25 km based on the background field of ERA5, and a daily dataset for 1961-1980 with a horizontal resolution of 10 km based on the background field of ERA5-LAND. Historical observations were bias-corrected and assimilated into the background fields to maximize local representativeness, especially in data-poor regions. In general, evaluation against independent station data shows that VnHGC substantially reduces systematic biases and better resolves regional patterns of temperature and precipitation, with the greatest improvements in northern and mountainous areas where reanalyses typically underperform. These improvements are especially pronounced in the daily dataset (1961-1980), reflecting the combined effects of higher temporal resolution and greater station availability