UAV-based Data collection / Kaddour Messaoudi

Public ISBD Titre : UAV-based Data collection Type de document : document multimédia Auteurs : Kaddour Messaoudi, Auteur ; Taher Bendouma, Directeur de thèse ; Omar Sami Oubbati, Directeur de thèse Editeur : Laghouat : Université Amar Telidji - Département d'informatique Année de publication : 2025 Importance : 147 p. Accompagnement : 1 disque optique numérique (CD-ROM) Langues : Anglais Mots-clés : Data collection  UAVs  Age of information (AoI)  UGVs  Reinforcement learning Résumé : The Internet of Things (IoT) continuously generates immense amounts of data, requiring efficient collection and processing. The unique features of IoT devices, such as limited energy, large-scale deployment, and potential mobility, pose numerous challenges to the optimized data collection process. Unmanned Aerial Vehicles (UAVs), commonly called drones, have emerged as a promising solution due to their mobility, ability to mitigate IoT communication limitations, minimize data collection delays, power IoT devices, and efficiently deliver data. However, UAVs are considered energy-constrained devices with limited computational and communication capacities. In this context, we are interested in handling these challenges by integrating UAVs as data collectors and energy transmitters. UAVs are supported from the ground by other intelligent devices, commonly known as Unmanned Ground Vehicles (UGVs), which serve as mobile charging stations. The main objective behind this thesis is to minimize the Age of Information (AoI) of the collected data while ensuring timely recharge and trajectory optimization for both UAVs and UGVs. Therefore, some solutions are proposed as main contributions using AIbased multi-agent Deep Reinforcement Learning (DRL) methods. First, we designed a novel scheme that involves the deployment of a single UAV to act as a data collector from a set of terrestrial IoT devices randomly distributed in a large-scale area. To this end, we leverage a single UGV dispatched on the ground to follow the UAV and charge whenever needed. Second, we introduced another UAV-UGV-based system of synchronized UAVs and UGVs to collaboratively perform the data collection process in a timely, efficient, and equitable manner while minimizing the AoI in a large-scale IoT environment. Finally, all proposed solutions, validated through extensive simulations, demonstrate significant improvements in data collection freshness and energy efficiency. note de thèses : Thèse de doctorat en informatique UAV-based Data collection [document multimédia] / Kaddour Messaoudi, Auteur ; Taher Bendouma, Directeur de thèse ; Omar Sami Oubbati, Directeur de thèse . - Laghouat : Université Amar Telidji - Département d'informatique, 2025 . - 147 p. + 1 disque optique numérique (CD-ROM). Langues : Anglais Mots-clés : Data collection  UAVs  Age of information (AoI)  UGVs  Reinforcement learning Résumé : The Internet of Things (IoT) continuously generates immense amounts of data, requiring efficient collection and processing. The unique features of IoT devices, such as limited energy, large-scale deployment, and potential mobility, pose numerous challenges to the optimized data collection process. Unmanned Aerial Vehicles (UAVs), commonly called drones, have emerged as a promising solution due to their mobility, ability to mitigate IoT communication limitations, minimize data collection delays, power IoT devices, and efficiently deliver data. However, UAVs are considered energy-constrained devices with limited computational and communication capacities. In this context, we are interested in handling these challenges by integrating UAVs as data collectors and energy transmitters. UAVs are supported from the ground by other intelligent devices, commonly known as Unmanned Ground Vehicles (UGVs), which serve as mobile charging stations. The main objective behind this thesis is to minimize the Age of Information (AoI) of the collected data while ensuring timely recharge and trajectory optimization for both UAVs and UGVs. Therefore, some solutions are proposed as main contributions using AIbased multi-agent Deep Reinforcement Learning (DRL) methods. First, we designed a novel scheme that involves the deployment of a single UAV to act as a data collector from a set of terrestrial IoT devices randomly distributed in a large-scale area. To this end, we leverage a single UGV dispatched on the ground to follow the UAV and charge whenever needed. Second, we introduced another UAV-UGV-based system of synchronized UAVs and UGVs to collaboratively perform the data collection process in a timely, efficient, and equitable manner while minimizing the AoI in a large-scale IoT environment. Finally, all proposed solutions, validated through extensive simulations, demonstrate significant improvements in data collection freshness and energy efficiency. note de thèses : Thèse de doctorat en informatique

---

page 1/1