A New Architecture of Wi-Fi APs with Prediction-based Autonomous Channel Migration Function

Wi-Fi throughput enhancement is required for the Tokyo Olympic in 2020. This is mainly because Wi-Fi can be a monetary-cost-effective solution for foreign tourists. However, especially in urban areas, Wi-Fi access points (APs) have deployed densely for the coverage enhancement of each service. Due to the dense deployment, there is a chaotic and disorderly environment in urban cities. Chaotic frequency usage causes interference and performance degradation. Channel migration is a simple and effective method for alleviating interference. In 2.4GHz band, due to the partial channel overlapping, some transmitted power leaks to the adjacent channels and becomes noise. This makes it more complex to select the best channel considering the traffic and signal strength diversity. For such a dense Wi-Fi problem, the IEEE802.11ax task group reports that Wi-Fi throughput can be nearly doubled [3]. Meanwhile, our goal is autonomous and efficient frequency reuse of each AP which has the existing architecture like IEEE802.11g. In our previous work, we have designed a function formula that predicts the Wi-Fi performance at each Wi-Fi AP, based on the measurement of IEEE802.11 MAC frames, the highly-precise Wi-Fi simulation, and machine learning technique [2]. In this paper, we propose a new architecture for intelligent APs that incorporate the prediction function formula.