Visualizing Wi-Fi Access Point measurements and location data using graph layouts

Abstract

Building and maintaining a database of Wi-Fi measurements in order to enable global WLAN-based positioning is a challenging and costly task, one whose cost can be minimized by crowdsourcing measurements from mobile phone users. Nonetheless, crowdsourcing Wi-Fi measurements poses a number of challenges, including the sheer size of data one must process and manage. In this paper, we present methods aimed at visualizing a large dataset of Wi-Fi signal strength measurements (n ≈ 2.3 × 106). Some of the measurements include associated location data, whereas a large number do not. This situation led us to represent the data as a graph, capturing the inherent spatial structure of Wi-Fi scan data. We define a subset of Access Points (APs) as “anchor nodes” and estimate their positions using a generalized least squares (GLS) estimator. We then employ a force-directed graph layout algorithm to lay-out the remaining APs in relation to the anchor nodes. The resulting graph layout provides a visualization of the entire dataset, which is inherently useful for further investigations, such as identifying where future data collection campaigns should be targeted.