Bluetooth Distance Estimation for COVID-19 Contact Tracing

Abstract

Because of the covid-19 pandemic, Bluetooth is widely adopted for contact tracing Apps to keep and prove social distancing. If two persons are close at a short distance as defined for a period of usually at least 15 minutes, then the contact should be automatically detected using Bluetooth Low Energy (BLE) measurements on the mobile devices of the two persons. For that purpose, usually the signal strength of the Bluetooth signals, referred to as Received Signal Strength Indicator (RSSI), is measured and converted into a distance using path loss models. Logarithmic models are thereby commonly employed. In this study, the feasibility of the use of BLE for this type of application is investigated. A test field in an indoor environment has been defined and measurements taken with different smartphones serving either as signal broadcaster, the so-called advertisers, or as scanners recording the BLE signals from the advertisers. From the RSSI measurements, distances are estimated and aerial distributions in the form of interpolated radio maps (or heat maps) derived. Experiments were conducted in three scenarios where the smartphones were either placed unobstructed in free space on chairs, put into backpacks or handbags and into the trousers pockets of the users. The results indicate that a meaningful relationship between the RSSI values and models based on an approximation with a logarithmic path loss model can be derived in most cases especially at a very close range (> 1 m). This is very promising if we consider the contact tracing application. From the radio maps of the whole test area, it could be seen that the results of the distribution of RSSI in the main free space and backpack experiments were coherent to the distance from each selected advertiser. The results of the trousers pocket experiment, however, showed unexpected distributions due to the low granularity in the sampling points.