Maximum Throughput of IEEE 802.11 Access Points: Test Procedure and Measurements

Performance measurements of IEEE 802.11 access points are becoming increasingly important since it is the predominant choice for high-speed wireless LAN networks. Among the different possible access point’s figures of merit, we focus our investigation on throughput performance. In particular, in this work we present measurements of the access point’s maximum saturation throughput. The saturation throughput is achieved when the access point always has a frame ready to transmit, and it reaches the maximum for optimal transmission conditions and appropriate offered traffic. We show that the maximum saturation throughput is the key figure of merit to characterize the performance of the access points. Several different standards are currently part of the 802.11 family, but we restrict our investigation to IEEE 802.11b devices. This standard is currently the most mature, with the largest number of devices, and with the most complete set of tools. Nevertheless, since the newer 802.11a and 802.11g are rapidly increasing their importance, we designed our measurement method in a way that it is possible to use it with others 802.11 technologies. The first contribution of this work is a methodology to produce and measure the maximum saturation throughput of any 802.11b access point. The proposed procedure includes the testbed setup, the software tools and the mathematical support for processing the results. All these aspects were investigated in order to define a unique and repeatable test procedure. The main effort was invested maximising and stabilising the access point performance. It was necessary to limit the impact of external factors that interfered with the access point optimal status and produced large throughput instability. We use the average value of the maximum saturation throughput as the figure of merit for 802.11 APs, and the test procedure allowed computing such a value with a defined confidence level and accuracy. The instant value of the saturation throughput even in optimal conditions has a too large variance to be significant. The second contribution of this work is the measurement results obtained following our proposed methodology from five different IEEE 802.11b access points. The purpose of the tests was to validate our methodology, and this result was successfully achieved. Moreover, the analysis of the test results suggests some conclusions about the performance of current access points. The downlink maximum saturation throughput is often limited by the access point performance, while the uplink maximum saturation throughput is rarely compromised. Therefore, the downlink saturation throughput is the key figure of merit of the access point, and the generic maximum saturation throughput of any access point is the 802.11 link capacity. On the other hand, increasing the offered load to the access point’s Ethernet interface does not always increase the downlink throughput; some access points present a downlink throughput reduction when the offered load exceeds their bridging capabilities. Finally, despite all the access points claim to use omni-directional antennas, some of them exhibit better performance