|dc.description.abstract||IEEE 802.11 standard has evolved from the early basic transmission rate to today's multiple transmission rates. The performance of IEEE 802.11 devices has been improved by exploiting multiple transmission rates. However, to take advantage of such evolution, a mechanism is required to choose the most appropriate rate under different circumstances, namely rate adaptation.
Rate adaptation is critical for improving system performance by exploiting multiple transmission rates provided by the current IEEE 802.11 WLANs (Wireless LANs) and adjusting the data rate accordingly under different channel conditions. The key challenge in designing such an algorithm is to select the most suitable data rate under different environments in order to maximize the throughput over wireless links. For a given channel, the poorer the channel quality, the lower should the data rate be. However, in a congestion dominated network, lowering the data rate does not help the situation, instead, it further decreases throughput because lower transmission rate will increase both transmission range and transmission time and therefore introduces more collisions.
Multiple rate adaptation schemes for IEEE 802.11 WLANs have been proposed and studied. The first generation rate adaptation schemes work well in collision free environments. However, in a congestion dominated environment, these schemes perform poorly because they do not differentiate frame losses caused by collision from those frame losses caused by channel degradation and unnecessarily decrease the data rate. The second generation rate adaptation schemes use Request To Send/Clear To Send (RTS/CTS) control frames to differentiate frame losses. However, there exist several problems when using these control frames. One obvious problem is the introduction of overhead which may lower network performance especially when the data frame size is small. The second problem is that, in IEEE 802.11 standard, the RTS is always transmitted at the lowest rate which may waste bandwidth under certain circumstances. The third problem is deciding when to turn on and off RTS/CTS control frames to reduce the overhead.
Besides the problems mentioned above, most current rate adaptation schemes only consider the situation when the channel quality is good or when there are a lot of collision. Little or no action has been taken when the signal strength is low. According to our study, special actions need to be taken when the channel quality is poor. In our proposed algorithms, we have made several adjustments to accommodate this situation.
This dissertation first gives several guidelines on how to design an efficient rate adaptation scheme and then presents two practical rate adaptation algorithms called Advanced Rate Adaptation (ARA) and Fast Recovery Rate Adaptation (FRRA). These two algorithms fully satisfy the proposed guidelines and are implemented along with four other representative rate adaptation schemes on a Linux-based testbed. The proposed algorithms and other selected rate adaptation schemes are evaluated extensively in both controlled and public field tests. Experiment results show that ARA and FRRA outperform other rate adaptation schemes in most scenarios.||en