Performance enhancement in wireless local area networks.

Abstract

In recent years, the IEEE 802.11 WLANs (Wireless Local-Area Networks) have received significant attention due to their higher bandwidth than wide-area cellular systems and their use of unlicensed (free-to-use) operational bands. Most 802.11-compliant products available in the market only implement the mandatory DCF (Distributed Coordination Function) at the MAC (Medium Access Control) layer, and because of the inherent design flaws of the DCF protocol, the current 802.11 systems present poor fairness-performance/energy-efficiency/channel-utilization. Now, with the emergence of new high-speed 802.11 PHYs (physical layers), the problems become even worse, and it is essential to re-examine the original DCF and make necessary modifications. This dissertation addresses the problem of enhancing the performance of the 802.11 DCF systems from various related but distinct angles. To achieve the weighted fairness and maximize the channel utilization for data communications in 802.11 DCF systems, we propose a simple enhancement to the 802.11 DCF, called the WB-DCF. It reduces the number of contending stations by introducing a new polling mode in addition to the contention mode used in the DCF, and achieves a low frame collision probability with an advanced contention window selection scheme based on runtime load estimation. Besides, the relative weights of traffic flows are also taken into consideration in the polling scheme and the contention window selection to achieve the weighted fairness. We investigate the problem of minimizing the energy consumption in the emerging 802.11a/h systems that will provide a structured means to support intelligent TPC (Transmit Power Control). We develop a novel scheme, called MiSer, as an optimal solution. The key idea is to combine TPC with PHY rate adaptation and compute offline an optimal rate-power combination table, and then at runtime, a wireless station determines the most energy-efficient transmission strategy for each data frame transmission by a simple table lookup. Using a similar table-driven idea, we also develop an intelligent link adaptation scheme, called ILA, for 802.11a DCF systems, which fully exploits the multiple transmission rates of the 802.11a PHY. ILA is able to react properly and quickly to the dynamically-changing network conditions and select the most appropriate transmission rate for the next transmission attempt. Finally, we implement a new RT-WLAN device driver module, which extends the original Linux device driver for Agere ORiNOCO cards to support soft real-time communications. RT-WLAN uses separate queues for real-time and non-real-time traffic and the service preference is given to the real-time queue. By serving the real-time queue according to the EDF (Earliest-Deadline-First) policy and using an adaptive traffic smoother to regulate bursty non-real-time traffic, the desired real-time support and service differentiation among real-time sessions are achieved with RT-WLAN.