Exploring the Usage of BLE in Crowd Environments

Abstract

Abstract Bluetooth Low Energy (BLE) is a pivotal low-cost technology for the Internet of Things (IoT). In recent years, BLE becomes widely used for short-range proximity and location estimation. With billions of people carrying BLE-equipped smartphones, many applications are already developed, based on BLE technology, to provide these people with different services according to their proximity and location. Nevertheless, the existence of human bodies, which constitute 80% water, is proved to influence BLE signals. This study targets the case of taking the people's crowdedness to its maximum extreme. Consequently, it studies and quantifies the effect of this very high crowdedness on the performance of BLE-based positioning systems. To present a truthful practical evaluation of the effect of this overcrowdedness, it conducts extensive real experiments in one area that witnesses too much crowd all over the year, the Holy Prophet’s mosque in Madinah, Saudi Arabia. The experiments are done and the data are collected after prayers, where the crowd density typically reaches 4 persons/𝑚2. The experiments evaluate both proximity and location beacons with different transmission powers. For both types of beacons, it evaluates the effect of overcrowdedness on the received signal strength, the coverage range, and the accuracy of proximity and location estimation processes. Furthermore, it validates the suitability of using the logarithmic path loss model within very crowded environments and finds its parameters. Compared to the uncrowded environments, results show random, but often manageable, changes in the received signal strength and estimation accuracy. Taking location beacons as examples, in comparison to uncrowded environments, the overcrowdedness changes the signal strength by up to 8.6 dBm, the coverage range by up to 24m, the proximity estimation accuracy by up to 18%, and the location estimation accuracy by up to 41%. Furthermore, the overcrowdedness causes the signal strength to fluctuate at the same position by up to 7%. Finally, the logarithmic path loss model is shown to be suitably used within very crowded environments with an R2 goodness-of-fit metric that is greater than 80%. The observations provided in this work can provide a reference for future researchers in terms of how well iBeacon estimote devices perform in overcrowded environments.