WiNetSense: Sensing and Analysis Model for Large-scale Wireless Networks

The enormous growth of smart devices, applications have moved Wireless Fidelity (WiFi) from a nascent stage to a widely accepted technology. However, WiFi has not been able to use its full potential because of various reasons such as low range, dynamic environment, shared spectrum, distributed design, and dense deployments. Therefore, there is a need to make WiFi network more reliable, effective, and manageable. Machine Learning (ML) algorithms are making way to utilize large volumes of data generated to monitor and manage large wireless networks. In this paper, we propose data driven sensing and analysis framework for making WiFi networks efficiently manageable. Our proposed framework referred as WiNetSense can be used to monitor and analyze the network features that are affecting the wireless links and also the network. We also propose a novel mobility estimation method using WiNetSense, which requires as minimum as two Access Points (APs) only to estimate the WiFi station's (STA) position and mobility. This knowledge can be used to trigger end user specific decisions such as smooth hand-off, load balancing, efficient mobility management, and dynamic power control.