Characterizing Bluetooth Low Energy Beacons for Studying Social Behavior

Bluetooth low energy (BLE) beacons are a relatively new technology enabled by both the rise of cheaper sensors for the Internet of Things (IoT), and innovations in the Bluetooth standard to decrease energy cost. As a result, BLE beacons have continued to decline in price and can now last several years on readily available commercial batteries, allowing their use in a variety of applications such as indoor location, occupancy detection, building utilization, and others. Meanwhile, IoT innovations have also enabled an increase in available sensors on commercial smartphones, as well as fueled their incredible popularity, causing not only technology researchers but those from medical, psychology, sociology, economics, and other areas to appreciate the value of smartphones for large-scale, unobtrusive, behavioral research. Of particular interest to many is the study of social behavior, which can consist of communication, co-location, face-to-face conversations, and more. My research lies in the center of these three movements, characterizing and using BLE beacons, in conjunction with the smartphones that detect them, for the large-scale study of social behavior.

In this work, I evaluate the use of BLE beacons for sensing social behavior in three ways. First, I will show that direct distance estimation from the received signal strength indicator (RSSI) of BLE, commonly used for classic Bluetooth in previous social sensing work, is difficult due to high signal variability at farther distances, especially when considering real-world sensing scenarios such as carrying the smartphone in a pocket. Second, I will present a new fingerprinting method for room-level indoor positioning that relies on which beacons are detected, not the RSSI reading, yet has high performance and robustness in real-world sensing situations on-par with or better than some traditional RSSI methods. Finally, I will discuss two case studies, room utilization and friendship prediction from dining visit habits alone, using data collected from a 40-beacon deployment in a campus dining hall, containing nearly 200 users from a sizeable behavioral sensing study, NetHealth.