Towards Optimal Access Point Selection with Available Bandwidth Estimation

Abstract

To achieve ubiquitous connectivity, the Internet of Things utilize WiFi as one of key wireless technologies. With densely deployment, the coverages of different WiFi access points (APs) are inevitably overlapped. Such overlaps lead to the AP selection problem. Traditionally, AP selection is based on received signal strength indication (RSSI). However, RSSI cannot reflect the performance of AP or network. To overcome this shortcoming, many researches take each AP's available or residual bandwidth into account. But their available bandwidth estimations are rough and inaccurate. Further, many of them require some modifications of users. In this paper, we first focus on estimating available bandwidth via analyzing communication process between users and APs. We then construct an effective and centralized AP selection strategy which allows IoT users seated in the overlapped area to dynamically connect to the best AP. All AP selection processes are executed by the AP and Local controller without any modification on the users. We verify our work via numerical studies and simulations. The results show that our available bandwidth estimation is accurate, and the proposed AP selection strategy performs well. The total network's throughput could be enhanced by up to 49% comparing with RSSI-based AP selection.