BEGIN:VCALENDAR VERSION:2.0 PRODID:-//wp-events-plugin.com//7.4.5//EN TZID:Europe/Paris X-WR-TIMEZONE:Europe/Paris BEGIN:VEVENT UID:0-296@lisn.upsaclay.fr DTSTART;TZID=Europe/Paris:20250917T090000 DTEND;TZID=Europe/Paris:20250917T090000 DTSTAMP:20250910T192646Z URL:https://www.lisn.upsaclay.fr/evenements/reinforcement-learning-and-fed erated-learning-based-multi-band-assignment-for-iot-short-packet-communica tions/ SUMMARY:Reinforcement Learning and Federated Learning-based Multi-Band Assi gnment for IoT Short Packet Communications DESCRIPTION:Jury members\nAndré-Luc BEYLOT\, Professor\, Toulouse INP\, Re viewer &\; Examiner \nKandaraj PIAMRAT\, Maitre de Conferences\, Nantes Université\, Reviewer &\; Examiner \nNadjib AIT-SAADI\, Professor\, U niversité Paris-Saclay\, Examiner \nFatiha ZAIDI\, Professor\, Universit é Paris-Saclay\, Examiner\nSteven MARTIN\, Professor\, Université Paris- Saclay\, Examiner \nYusheng JI\, Professor\, National Institute of Informa tics\, Examiner\nKensuke FUKUDA\, Professor\, National Institute of Inform atics\, Examiner \nTakashi KURIMOTO\, Professor\, National Institute of In formatics\, Examiner \nAbstract\n5G services are classified into Ultra-Rel iable Low-Latency Communications (URLLC)\, enhanced Mobile Broadband commu nications (eMBB)\, and massive Machine Type Communications (mMTC). Beyond 5G (B5G) and 6G communications are expected to cover extreme demands\, joi ntly in terms of rate\, delay\, reliability. Emergent applications such as Extended Reality (XR) or autonomous driving are expected to require Terab its data rate\, extreme reliability and/or delay. Moreover\, given the tre mendous amount of wireless devices\, and especially IoT devices\, meeting the requirements of future networks appears to be extremely challenging. M any recent studies have focused on the use of the high millimeter Wave (mm Wave) frequencies due to their large bandwidth. However\, mmWaves suffer o f a severe sensitivity to obstacles and a high path loss\, restricting the ir applicability. To overcome these frailties\, many works proposed to joi ntly use the conventional Sub-6GHz frequencies and the mmWave. Most IoT ap plications\, for example in industrial control\, can be characterized by t heir generation of small packets (Short-Packet Communications (SPC)). For such traffic\, Shannon’s capacity theorem becomes inapplicable due to th e size and nature of SPC. To better model the achievable rates\, we adopt the Finite Blocklength theory\, which accounts for packet size and reliabi lity constraints. This thesis investigates the problem of band association and radio resource allocation of IoT devices to maximize the system sum-r ate under heterogeneous Quality of Service (QoS) demands. We first propose a unifying framework devoted to SPC that jointly optimizes the user parti tioning\, and the radio resource scheduling within each band. Leveraging c entralized Deep Reinforcement Learning (DRL)\, the proposed method enables to better tackle the challenges imposed by dynamically varying mobile env ironments. Regarding the DRL-based user partitioning\, we have investigate d three different types of actions to obtain a high network performance an d rapid convergence. Regarding the resource allocation\, we designed two o ptimization methods\, one that leverages Difference of Convex Programming (DCP)\, and the second that accelerates convergence. Numerical evaluations show that the proposed methods outperform conventional approaches. While our centralized framework efficiently optimizes users' band associations\, centralization entails important communication costs and increasing compl exity. Recently\, Multi-Agent DRL (MADRL) approaches attracted attention\, as solving the problem at the edge avoids communication burden. However\, efficient solutions must be provided to allow efficient coordination of d istributed devices. To enhance coordination while avoiding heavy communica tion costs\, Federated MADRL (F-MADRL) has been studied where each device trains its model locally while a periodic model aggregation done at a cent ral node improves cooperation. Since each IoT device faces a unique mobile environment\, personalized F-MADRL was introduced to improve local decisi on making. We propose two personalized F-MADRL methods that enable each us er to adapt their model to the mobile environment\, thereby improving the overall network performance and reducing signaling costs. Simulations show that our methods outperform benchmark MADRL schemes. Finally\, we focus o ur work on denser and more dynamic network scenarios. We evaluate a third personalization strategy to improve robustness in heterogeneous environmen ts. To reduce the complexity of the DCP-based resource allocation\, we exp lore low-complexity algorithm\, as the mathematical optimization we used u ntil now requires a computationally intensive process. Results show promis ing performance of our proposed solution in denser networks\, outperformin g conventional algorithm in terms of outage probabilities or sum-rate. Fin ally\, we evaluate robustness under user mobility and blockage\, showing t he fast adaptivity of our personalized F-MADRL framework. CATEGORIES:Science des Données,Thèses et HDR END:VEVENT BEGIN:VTIMEZONE TZID:Europe/Paris X-LIC-LOCATION:Europe/Paris BEGIN:DAYLIGHT DTSTART:20250330T030000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST END:DAYLIGHT END:VTIMEZONE END:VCALENDAR