Research On The AP Performance Enhancement Techniques In WLAN

With the popularity of mobile terminals, the tide of Mobile Internet is flooding everywhere. The wireless LANs are the connector between the Internet and mobile data communication on terminals. As a high-speed wireless broadband access method, WLANs have its unique advantages in terms of high-bandwidth, low-cost, fast and flexible networking and etc. Wireless LANs have become an important way for people to access the Internet in their daily lives, and they will still account for a growing proportion.When the wireless LANs are acting as access network, most of them are infrastructure WLANs. Any infrastructure WLAN which is based on DCF mechanism or derived from CSAM/CA mechanism would meet a problem that the channel resources allocated to Access Point and Stations does not match the traffic loaded on them. It is called uplink/downlink asymmetry in this paper. Specifically, downlink traffic is more than uplink traffic in most case, but Access Point and one STA has equal probability to access the channel. So downlink data is easy to heap up in MAC queue of Access Point which cause drop of downlink throughput, rise of end-toend delay and queue tail packet loss. The solution presented in this paper is called APC(AP Priority Control),which changes the access parameters of AP adaptively(interframe space,contention window and TXOPlimit). The purpose is to allow the WLAN fit in the asymmetric uplink/downlink services.In this thesis we adopt a three-dimensional Markov chain to analyse the performance of APC. By comparison with NS3 network simulaton, it is proved that the mathematical model is relatively accurate. With the model, this paper found parameters of APC when the downstream to the upstream traffic load is about 4:1, and did two sets of experiments on NS3 platform to show the effect of APC compared with control group without APC.The first set of experiments shows the changes of uplink/downlink throughput, end-to-end delay and queue tail-drop rate with different arrival rates of traffic. By comparison with the control group, we can draw the following conclusions: 1, APC can effectively enhance the downlink throughput while reduce the uplink throughput. No matter how much traffic is loaded, APC always makes downlink throughput four times the uplink throughput which fit in theexperimental scenarios(downlink traffic amount is four times the uplink traffic). So the total throughput is improved by 44%. 2, APC also effectively reduces the end-to-end delay of downlink data which doesn’t change much with different arrival rates of traffic. But APC increases the uplink data end to end delay of about 1 millisecond. 3, Access Point has equal probability to access the channel with one STA but downlink traffic is much more than uplink traffic in control group, so downlink queue tail-drop rate is much more than uplink’s. With the introduction of APC, the queue tail-drop rate of uplink and downlink tends to be the same.The second set of experiments shows the changes of uplink/downlink throughput, end-to-end delay and queue tail-drop rate with different number of STAs. By comparison with the control group, we can draw the following conclusions: 1,with the increase in the number of STAs, downlink data throughput and uplink data throughput increases linearly. When the number of STA changes from 4 to 8, the uplink data throughput increases greater than 2 times, while the increase of the downlink data throughput is less than 2 times, but the downlink data is always higher than the uplink data. But in the control group, when the STA number reaches 8, the uplink data is over the downlink data. 2, since the introduction of APC, increase in the number of STAs effects little on the downlink end-to-end delay. With the increase in the number of STAs, downlink data end-to-end delay increases by 8 microseconds, and the mathematical model increases by 50 microseconds, but the end-to-end delay of the uplink data are grows rapidly.3, although the end-to-end delay of AP is small, but with the increase of STAs, the downward load increases so that the data pours into the AP’s MAC queue until it overflows. The downlink queue tail-drop rate is greater than upward one, but compared with the control group, APC reduces the downlink queue tail-drop rate, increases the uplink’s, so that the gap between the two is small because of APC.