<p>IEEE 802.11 technology provides a low-cost wireless networking solution. In the last few years, we have seen that the demand for high-bandwidth wireless local area networks increases rapidly, due to the proliferation of mobile devices such as laptops, smart phones and tablet PCs. This has driven the widespread deployment of IEEE 802.11 wireless networks to provide Internet access. However, wireless networks present their own unique problems. Wireless channel is extremely variable and can be affected by a number of different factors, such as collisions, multipath fading and signal attenuation. As such, rate adaptation algorithm is a key component of IEEE 802.11 standard which is used to vary the transmission data rate to match the wireless channel conditions, in order to achieve the best possible performance. Rate adaptation algorithm studies and evaluations are always hot research topics. However, despite its popularity, little work has been done on evaluating the performance of rate adaptation algorithms by comparing the throughput of the algorithm with the throughput of the fixed rates. This thesis presents an experimental study that compares the performance ofMikroTik rate adaptation algorithm andMinstrel rate adaptation algorithm against fixed rates in an IEEE 802.11g network. MikroTik and Minstrel rate adaptation algorithm are most commonly used algorithm around the world. All experiments are conducted in a real world environment in this thesis. In a real world environment, wireless channel conditions are not tightly being controlled, and it is extremely vulnerable to interference of surrounding environment. The dynamic changes of wireless channel conditions have a considerable effect on the performance of rate adaptation algorithms. The main challenge of evaluating a rate adaptation algorithm in a real world environment is getting different experiment behaviours from the same experiment. Experiment results may indicate many different behaviours which due to the leak of wireless environment controlling. Having a final conclusion from those experiment results can be a challenge task. In order to perform a comprehensive rate adaptation algorithm evaluation. All experiments run 20 times for 60 seconds. The average result and stand deviation is calculated. We also design and implement an automation experiment controlling program to help us maintain that each run of experiment is following exactly the same procedures. In MikroTik rate adaptation algorithm evaluation, the results show in many cases that fixed rate outperforms rate adaptation. Our findings raise questions regarding the suitability of the adopted rate adaptation algorithm in typical indoor environments. Furthermore, our study indicates that it is not wise to simply ignore fixed rate. A fine selection of a fixed rate could be made to achieve desired performance. The result ofMinstrel rate adaptation evaluation show that whilst Minstrel performs reasonably well in static wireless channel conditions, in some cases the algorithm has difficulty selecting the optimal data rate in the presence of dynamic channel conditions. In addition, Minstrel performs well when the channel condition improves frombad quality to good quality. However, Minstrel has trouble selecting the optimal rate when the channel condition deteriorates from good quality to bad quality. By comparing the experimental results between the performance of rate adaptation algorithms and the performance of fixed data rate against different factors, the experiment results directly pointed out the weakness of these two rate adaptation algorithms. Our findings from both experiments provide useful information on the design of rate adaptation algorithms.</p>
Contextual Bandit Based Rate Adaptation Mechanism for IEEE 802.11 Mobile Wireless LANs
