ETD

Archivio digitale delle tesi discusse presso l'Università di Pisa

Tesi etd-09252007-100435


Tipo di tesi
Tesi di dottorato di ricerca
Autore
TAVANTI, LUCA
Indirizzo email
luctav@interfree.it
URN
etd-09252007-100435
Titolo
Traffic Control and Quality of Service in Wireless LANs
Settore scientifico disciplinare
ING-INF/03
Corso di studi
INGEGNERIA DELL'INFORMAZIONE
Relatori
Relatore Prof. Russo, Franco
Relatore Dott. Garroppo, Rosario Giuseppe
Relatore Prof. Giordano, Stefano
Parole chiave
  • Traffic Control
  • Scheduler
  • Quality of Service (QoS)
  • Proportional Fairness
  • IEEE 802.11
  • Admission Control
  • Wireless LANs
Data inizio appello
25/05/2007
Consultabilità
Completa
Riassunto
The thesis deals with two aspects of the IEEE 802.11 standard. The first is the so-called “performance anomaly”: the variable bandwidth of the links and the use of multiple transmission rates push the throughput of all stations to align to the slowest one. To tackle this problem we designed and developed a simple channel-aware scheduling algorithm, called DTT, which actualises the proportional fairness concept, thus leading to noteworthy improvements, and in particular to flow isolation. This is achieved by measuring link quality as the time needed to deliver a frame. The resource to share is no longer capacity, but the time the channel is in use. DTT has then been integrated into a prototype Access Point, which is the first working implementation of a scheduler based on proportional fairness.
Secondly, we focused on 802.11e networks, which, though enhancing QoS support, still offer scarce reliability of QoS guarantees and suffer from network congestion. We devised two admission control algorithms to assess the maximum number of users allowable to the services while satisfying QoS requirements. Following the studies on DTT, both algorithms centre the admission test on the time occupancy of the medium. The first algorithm builds on an analytical model of the EDCA mode in non-saturation conditions. This closely matches the real behaviour of a network carrying time-sensitive applications, thus overcoming the limits of all previous works, based on saturation models. The second algorithm uses and extends to 802.11e the NUC, a parameter defined and proved effective for 802.11b systems. This scheme needs measures of the actual state of the network. Simulations run within the E-model framework show good accuracy performance for both models.
File