In recent years, there have been tremendous advances in wireless & mobile communications, including wireless radio techniques, networking protocols, and mobile devices. It is expected that different broadband wireless access technologies, e.g., WiFi (IEEE 802.11) and WiMAX (IEEE 802.16) will coexist in the future. In the meantime, multimedia applications have experienced an explosive growth with increasing user demands. Nowadays, people expect to receive high-speed video, audio, voice and web services even when being mobile. The key question that needs to be answered, then, is how do we ensure that users always have the "best" network performance with the "lowest" costs in such complicated situations? The latest IEEE 802.11n standards attains rates of more than 100 Mbps by introducing innovative enhancements at the PHY and MAC layer, e.g. MIMO and Frame Aggregation, respectively. However, in this thesis we demonstrate that frame aggregation's performance adheres due to the EDCA scheduler's priority mechanism and consequently resulting in the network's poor overall performance. Short waiting times for high priority flows into the aggregation queue resolves to poor channel utilization. A Delayed Channel Access algorithm was designed to intentionally postpone the channel access procedure so that the number of packets in a formed frame can be increased and so will the network's overall performance. However, in some cases, the DCA algorithm has a negative impact on the applications that utilize the TCP protocol, especially the when small TCP window sizes are engaged. So, the TCP process starts to refrain from sending data due to delayed acknowledgements and the overall throughput drops. In this thesis, we address the above issues by firstly demonstrating the potential performance benefits of frame aggregation over the next generation wireless networks. The efficiency and behaviour of frame aggregation within a single queue, are mathematically analysed with the aid of a M=G[a;b]=1=K model. Results show that a trade-off choice has to be taken into account over minimizing the waiting time or maximizing utilization. We also point out that there isn't an optimum batch collection rule which can be assumed as generally valid but individual cases have to be considered separately. Secondly, we demonstrate through extensive simulations that by introducing a method, the DCA algorithm, which dynamically determines and adapts batch collections based upon the traffic's characteristics, QoS requirements and server's maximum capacity, also improves e ciency. Thirdly, it is important to understand the behaviour of the TCP ows over the WLAN and the influence that DCA has over the degrading performance of the TCP protocol. We investigate the cause of the problem and provide the foundations of designing and implementing possible solutions. Fourthly, we introduce two innovative proposals, one amendment and one extension to the original DCA algorithm, called Adaptive DCA and Selective DCA, respectively. Both solutions have been implemented in OPNET and extensive simulation runs over a wide set of scenarios show their effectiveness over the network's overall performance, each in its own way.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:569690 |
| Date | January 2013 |
| Creators | Skordoulis, Dionysios |
| Contributors | Ni, Q. |
| Publisher | Brunel University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://bura.brunel.ac.uk/handle/2438/7375 |
Page generated in 0.0017 seconds