WIOC: Wireless Indoor-Outdoor Classification Using WiFi and Cellular Signals

Today’s mobile devices are served by Heterogeneous Networks (HetNets) that consist of Wireless Fidelity (WiFi) access points (APs) often serving indoor areas, and cellular networks often serving large outdoor areas. The stand-alone nature of such network implementations leads to inter-network vertical handover (VHO), spectrum sharing, and interference management problems in regions of overlapping coverage. These challenges can be resolved by considering the user environment context to select the appropriate network, suitable spectrum, and acceptable power levels. In this paper, we address the problem of real-time user environment estimation and utilize it to improve wireless connection quality of experience (QOE). The WIOC system is designed to sense the variation patterns of signal strength values from nearby network nodes to determine the user environment. In contrast to earlier work, signal strength indicators from WiFi and Long Term Evolution (LTE) networks are parsed as time series data to an XGBOOST machine learning (ML) model trained to distinguish between indoor vs outdoor environments in real-time. The WIOC system achieves comparable classification accuracy to the current best method while offering improved generalization, making it well-suited for deployment in diverse scenarios. Finally, we demonstrate the utility of the WIOC framework in solving the sticky WiFi client problem (SWCP) by reducing the outdoor user connections to poor quality WiFi networks and thus reducing ping-pong inter-network handovers.