The physical layer protocols in the 802.11 wireless networks can transmit data at various transmission rates, and different transmission rates have different bandwidth, range and fading characteristics. MAC layer rate control mechanisms adaptively select transmission rates that will be used by the physical layer and such selections are based on an estimation of the current channel condition. Changes in channel conditions in the wireless networks can be quite dynamic and therefore the accurate evaluation of the current channel condition and the selection of an appropriate transmission rate has an impact on the network performance. The thesis addresses design, development and evaluation of MAC layer rate control mechanisms.
The thesis first presents a critical literature survey of existing rate control mechanisms, providing a classification of the rate control mechanisms based on the type of metrics used for evaluation of the channel conditions, the algorithms that they use for selecting transmission rates and a discussion on the suitability of these various metrics for the current wireless networks working with the Linux operating system.
The literature survey is followed by an evaluation of the four rate control mechanisms on the Madwifi driver in the Linux system, i.e. Minstrel, AMRR, Onoe and SampleRate. Minstrel is found to be the best rate control mechanism. As the mac80211 framework has emerged in the Linux system since kernel 2.6.22, the rate control mechanism is now a component of the framework instead of being dependent on individual drivers. The two rate control mechanisms available on the new mac80211 framework, Minstrel (ported from Madwifi) and PID, are compared and it is shown that Minstrel also outperforms PID. The reason for PID's poor performance is analysed and an enhancement is proposed to improve its performance.
Based on the experience gained in evaluation and improvement of rate control mechanisms, the thesis presents a new rate control mechanism, RCELC (rate control based on estimated link capacity). RCELC uses an accurate metric to estimate the link capacity and a robust algorithm to dynamically adapt the transmission rate in dynamic channel conditions. The new mechanism is evaluated in a range of conducted and over-the-air experiments and it is shown that it performs better than the Minstrel rate control mechanism.
Collision has a negative impact on the performance of rate control mechanisms. In this thesis, two types of collisions in wireless networks, contention and hidden terminal collisions, are discussed and their impact on rate control mechanisms is analysed, respectively. A collision-aware mechanism is also proposed to enhance rate control performance in hidden terminal environments.