logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-04072012-145024


Tipo di tesi
Tesi di dottorato di ricerca
Autore
MIGLIORINI, DANIELE
URN
etd-04072012-145024
Titolo
Resource Allocation in LTE Advanced for QoS and Energy Efficiency
Settore scientifico disciplinare
ING-INF/05
Corso di studi
INGEGNERIA DELL'INFORMAZIONE
Relatori
tutor Prof. Lenzini, Luciano
tutor Ing. Stea, Giovanni
Parole chiave
  • cellular networks
  • energy efficiency
  • LTE
  • resource allocation
  • scheduling
Data inizio appello
24/05/2012
Consultabilità
Completa
Riassunto
Long Term Evolution (LTE) and LTE-Advanced (LTE-A) are establishing themselves as the new standard of 4G cellular networks in Europe and in several other parts of the world. Their success will largely depend on their ability to support Quality of Service for different types of users, at reasonable costs. The quality of service will depend on how effectively the cell bandwidth is shared among the users. The cost will depend – among many other factors – on how effectively we exploit the cell capacity. Being able to exploit bandwidth efficiently postpones the time when network upgrades are required. On the other hand, operation costs also depend on the energy efficiency of the cellular network, which should avoid wasting power when few users are connected.
As for bandwidth efficiency, the recent LTE/LTE-A standards introduced MIMO (Multiple Input Multiple Output) transmission modes, which allow both reliability and efficiency to be increased. MIMO can increase the throughput significantly. In a MIMO system, the selection of the MIMO transmission modes (whether Transmission Diversity, Spatial Multiplexing, or Multi-User MIMO) plays a key feature in determining the achievable rate and the error probability experienced by the users. MIMO-unaware scheduling policies, which neglect the transmission mode selection problem, do not perform well under MIMO. In the current literature, few MIMO-aware LTE-A scheduling policies have been designed. However, despite being proposed for LTE-A, these solutions do not take into account some constraints inherent to LTE-A, hence leading to unfeasible allocations. In this work, we propose a new framework for Transmission Mode Selection and Frequency.
Domain Packet Scheduling, which is compliant with the constraints of the LTE-A standard. The resource allocation framework accommodates real-time requirements and fairness on demand, while the bulk of the resources are allocated in an opportunistic fashion, i.e. so as to maximize the cell throughput. Our results show that our proposal provides real-time connections with the desired level of QoS, without utterly sacrificing the cell throughput.
As far as energy efficiency is concerned, we studied the problem of minimizing the RF power used by the eNodeB, while maintaining the same level of service for the users. We devised a provisioning framework that exploits the Multicast/Broadcast over a Single Frequency Network (MBSFN) mechanism to deactivate the eNodeB on some Transmission Time Intervals (TTI), and computes the minimum-power activation required for guaranteeing a given level of service. Our results show that the provisioning framework is stable, and that it allows significant savings with respect to an always-on policy, with marginal impact on the latency experienced by the users.
File