Resource management for enabling heterogeneous services and applications in wireless cellular systems.

Nguyen, Ti Ti (2020). Resource management for enabling heterogeneous services and applications in wireless cellular systems. Thèse. Québec, Doctorat en télécommunications, Université du Québec, Institut national de la recherche scientifique, 240 p.

Prévisualisation

PDF Télécharger (2MB) | Prévisualisation

Résumé

5G New Radio (NR) and Mobile Edge Computing (MEC) have been recently proposed as important technologies and architectures for the next-generation wireless cellular system, which allows efficiently supporting emerging services and applications. Indeed, 5G NR provides a flexible frame structure with scalable Transmission Time Interval (TTI) in different so-called numerologies, which enable us to meet diverse quality of service (QoS) requirements of different wireless services and compute-intensive applications. However, many challenges must be resolved to efficiently utilize different types of system resources and better support heterogeneous wireless services. The overall objective of this Ph.D. research is to develop efficient resource management techniques for next-generation systems exploiting 5G NR and MEC technologies. Our research has resulted in three major research contributions, which are presented in three corresponding main chapters of this dissertation. First, we study fair computation offloading and resource allocation for the MIMO based MEC system, which is presented in Chapter 5. In particular, we formulate the joint computation offloading and resource allocation problem that minimizes the maximum weighted consumed energy for mobile users considering the latency and resource limitation constraints. Then, we propose different efficient algorithms to solve the underlying mixed-integer nonlinear programming (MINLP) problems under perfect and imperfect channel state information estimations. Second, we investigate the joint data compression, computation offloading, and resource allocation problem for hierarchical fog-cloud systems aiming to minimize the maximum weighted energy and service delay cost of all users, which is covered in Chapter 6. To this end, we propose a non-linear computation model which can be fitted to accurately capture the computational load incurred by data compression and decompression. Then, we first consider the scenario where data compression is performed only at the mobile users. A novel three-step approach using convexification techniques is developed to optimize the compression ratios and the resource allocation. As the next step, we address the more general design where data compression is performed at both the mobile users and the fog server. We propose three algorithms to overcome the strong coupling between the offloading decision and the resource allocation and find efficient solutions for the underlying problem.. Finally, we consider leveraging the 5G NR to support diverse applications with different requirements. The research outcomes of this study are presented in Chapter 7. In particular, we study the scheduling problem for heterogeneous services with mixed numerology which aims to maximize the number of admitted users while meeting service latency and data transmission requirements. Then, two algorithms, namely Resource Partitioning-based Algorithm (RPA) and Iterative Greedy Algorithm (IGA), are developed to acquire efficient resource scheduling solutions. For all the considered problems and proposed designs, extensive numerical results are presented to gain further insights and to evaluate the performance of our algorithms. Our numerical studies confirm that our algorithms can achieve efficient resource utilization, energy saving, and significant performance gains compared to existing designs.

Type de document:	Thèse Thèse
Directeur de mémoire/thèse:	Le, Long Bao
Mots-clés libres:	télécommunications
Centre:	Centre Énergie Matériaux Télécommunications
Date de dépôt:	14 oct. 2020 19:27
Dernière modification:	14 oct. 2020 19:27
URI:	https://espace.inrs.ca/id/eprint/10403

Gestion Actions (Identification requise)

Modifier la notice