Novel Wi-Fi Throughput Estimation Method Considering CSMA/CA Behavior

Abstract

Wi-Fi is one of the solutions to overcome the explosion of mobile data traffic and improvement of Wi-Fi performance intensifies the cellular data offload to Wi-Fi. Wi-Fi AP (Access Point) selection at the STA (STAtion) is a key function to improve Wi-Fi throughput. Our proposed method is practical for the AP selection. If an STA can know the throughputs between the STA and surrounding APs before exchanging Wi-Fi packets with an AP, the STA can select the optimum AP that promises high throughput. In this paper, we focus on a throughput estimation method that is applicable to AP selection, and discuss the accuracy of the estimated throughput. We propose a novel Wi-Fi throughput estimation method with passive measurement and without modification for the Wi-Fi standard. The proposed method consists of three stages: measurement of the time ratio used by each Wi-Fi device, selection of dominant Wi-Fi devices and calculation of estimated throughout. Our proposed method considers the influence of CSMA/CA that affects Wi-Fi throughput that changes based on the number of Wi-Fi devices sharing the same frequency band and the traffic load of each device. We evaluate the effectiveness of the proposed method by experiments (estimated before communication) and compare the estimated throughput and the actual throughput (measured during communication) in certain environments. The proposed method is effective for estimating Wi-Fi throughput. The RMSE (Root Mean Square Error) of the proposed method decreases to 29~33% that of conventional methods. Estimated throughput with our proposed method is always higher than measured throughput. Therefore, modification of the proposed method to decrease the difference between the estimated and measured value is a subject for future study. As some systems influence Wi-Fi, considering these influences in a throughput estimation process is also recommended.