Using Only 1 channel: Impact of Density

Figure 12: Benefits of density: DenseAP with 1 channel

We repeated the previously described experiment with only one channel, and we varied the number of active clients from one to six. We did the experiment for the corporate WLAN, and then repeated it for DenseAP with all DAPs set to use the same channel.

Note that association policy plays very little role in this setting. Our testbed is small, and all DAPs interfere with one another. As a result, load on all DAPs is the same, so the association policy is reduced to simply selecting a DAP that hears the client with reasonable signal strength. For similar reasons, load balancing does not play a role either.

Thus, the only factor providing gains for DenseAP in this setting is the density of the DAPs. The reason density provides performance gains in this setting is the following. As more clients are added, the performance of the corporate WLAN is dominated by the client with the worst connection quality, which is usually the client that is the farthest away from the AP. Due to poor connection quality, such clients use lower transmission rates, thereby consuming more airtime. This, in turn, hurts performance of all other clients. This is known as the rate anomaly problem [13]. With DenseAP however, each client generally talks to a nearby DAP. As a result, clients and DAPs can communicate at higher data rates, thereby reducing the impact of the rate anomaly problem.

The results of this experiment are shown in Figure 12. We see that DenseAP performs better than WLAN even in this setting. The results highlight the benefit of the dense AP deployment. They also explain why, in the previous section, we saw gains of more than 800% with 8 channels!

The results lead us to ask: can a system administrator significantly improve capacity by simply adding more APs to the network? In other words, what is the contribution of the association policy to the overall gain? We examine this in the next section.