logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-06082011-163158


Tipo di tesi
Tesi di dottorato di ricerca
Autore
ERMINI, DONATELLA
URN
etd-06082011-163158
Titolo
Traffic Scheduling in Point-to-Multipoint OFDMA-based Systems
Settore scientifico disciplinare
INF/01
Corso di studi
INFORMATICA
Relatori
tutor Prof. Bonuccelli, Maurizio
Parole chiave
  • Complexity
  • Heuristic Algorithms
  • OFDMA Wireless Systems
  • Real-time
  • Traffic Scheduling
Data inizio appello
21/06/2011
Consultabilità
Completa
Riassunto
The new generation of wireless networks (e.g., WiMAX, LTE-Advanced, Cognitive Radio) support many high resource-consuming services (e.g., VoIP, video conference, multiplayer interactive gaming, multimedia streaming, digital video broadcasting, mobile commerce). The main problem of such networks is that the bandwidth is limited, besides to be subject to fading process, and shared among multiple users. Therefore, a combination of sophisticated transmission techniques (e.g., OFDMA) and proper packet scheduling algorithms is necessary, in order to provide applications with suitable quality of service.
This Thesis addresses the problem of traffic scheduling in Point-to-Multipoint OFDMA-based systems. We formally prove that in such systems, even a simple scheduling problem of a Service Class at a time, is NP-complete, therefore, computationally intractable. An optimal solution is unfeasible in term of time, thus, fast and simple scheduling heuristics are needed. First, we address the Best Effort traffic scheduling issue, in a system adopting variable-length Frames, with the objective of producing a legal schedule (i.e., the one meeting all system constraints) of minimum length. Besides, we present fast and simple heuristics, which generate suboptimal solutions, and evaluate their performance in the average case, as in the worst one. Then, we investigate the scheduling of Real Time traffic, with the objective of meeting as many deadlines as possible, or equivalently, minimizing the packet drop ratio. Specifically, we propose two scheduling heuristics, which apply two different resource allocation mechanisms, and evaluate their average-case performance by means of a simulation experiment.
File