Dépôt numérique
RECHERCHER

Edge caching and network slicing for wireless cellular networks.

Téléchargements

Téléchargements par mois depuis la dernière année

Plus de statistiques...

Tran, Duy Thinh (2020). Edge caching and network slicing for wireless cellular networks. Thèse. Québec, Doctorat en télécommunications, Université du Québec, Institut national de la recherche scientifique, 199 p.

[thumbnail of Tran, Duy Thinh.pdf]
Prévisualisation
PDF
Télécharger (3MB) | Prévisualisation

Résumé

The fifth-generation (5G) wireless cellular system is expected to provide huge improvement in comparison to the fourth-generation (4G) system in supporting more stringent and versatile technical requirements. Particularly, the 5G system should be capable of providing a 1000-fold of network throughput, supporting ultra reliable and low latency communications, and handling massive connectivity. Novel techniques must be devised and well integrated to enable the 5G wireless cellular system fulfill such stringent key performance requirements of diverse wireless applications and to significantly reduce the capital expenditure (CAPEX) and operational expenditure (OPEX) for 5G cellular network operators. Wireless network virtualization (WNV), or network slicing, has been considered as a promising networking approach for addressing this problem. Efficient techniques for advanced resource management of network slices must be developed to achieve high resource utilization efficiency and flexibility while satisfying each slice’s quality of service (QoS) constraints. Harnessing new kind of resources (i.e., new resource dimensions) is essential to help the future 5G wireless cellular system satisfy these stringent technical requirements in addition to exploiting new spectrum bands (e.g., millimeter wave (mmWave)) and techniques for improving spectrum and energy efficiency. Moreover, content caching at the network edge, i.e, placing popular contents or files at places closer to end users can potentially help significantly reduce network traffic and access delay considering the rapid increase of mobile video data in the wireless cellular network. The 5G network performance would be further improved if different types of network resources such as frequency spectrum, transmission power, and storage resources were efficiently utilized and managed. As a result, it is crucial to design frameworks for network resource management and content placement. Motivated by these promising key directions, the general objective of this Ph.D. research is to develop efficient resource management techniques enabling wireless edge caching and network virtualization in the wireless cellular network. Our research has resulted in three major research contributions, which are presented in three corresponding chapters of this dissertation. First, we study the caching problem for heterogeneous small-cell networks with bandwidth allocation and caching-aware base station (BS) association. The caching control and bandwidth allocation problem aims at minimizing the request miss rate for one network operator who has a limited bandwidth and storage capacity in serving its end users (UEs). To solve this problem, we propose a Line-Search-based-Iterative (LSBI) algorithm which determines the solution by combining the line-search algorithm to obtain the optimal bandwidth allocation with the iterative caching algorithm to acquire a caching solution. Numerical results demonstrate that the LSBI algorithm significantly outperforms existing caching algorithms, and is on a par with a performance bound. Second, we investigate the joint resource allocation and content caching problem which aims to efficiently utilize the radio and content storage resources in multi-cell virtualized wireless network with highly congested backhaul links. In this design, we minimize the maximum content request rejection rate experienced by users of different mobile virtual network operators (MVNO) who share a common resource pool of subcarriers and storage repositories owned by an infrastructure provider (InP). We solve the resulting mixed-integer non-linear programming (MINLP) problem by proposing a bisection-search based algorithm that iteratively optimizes the resource allocation and content caching placement. We further propose a low-complexity heuristic algorithm which achieves moderate performance loss compared to the bisection-search based algorithm. Extensive numerical results confirm the efficacy of our proposed framework which significantly reduces the maximum request outage probability compared to other benchmark algorithms. Third, we study the resource allocation and pricing problem in the virtualized wireless network that captures the multilateral interactions among access/backhaul service providers and their UEs by using the multi-leader-multi-follower (MLMF) Stackelberg game approach. Toward this end, we show how to formulate such a Stackelberg game and prove the existence of a unique game equilibrium. Then, we develop a distributed algorithm based on updating underlying best-response functions, which is proved to converge to the game equilibrium. Numerical results are presented to provide important insights into the interactions among the involved stakeholders and demonstrate the economical efficacy of the proposed design with respect to existing benchmarks. In summary, different efficient resource management algorithms have been developed considering several enabling 5G wireless technologies. Moreover, extensive numerical results are presented in each contribution to gain further insights and to evaluate the performance of our proposed designs. The solid results achieved in this dissertation would form good foundations for our future studies where research issues such as mobility management, security and privacy, and applications of machine learning techniques for more effective network management can be addressed.

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 19:34
Dernière modification: 14 oct. 2020 19:34
URI: https://espace.inrs.ca/id/eprint/10416

Gestion Actions (Identification requise)

Modifier la notice Modifier la notice