Bluetooth Low Energy (BLE) Precision Tests¶
This document summarizes the results of the Bluetooth Low Energy (BLE) tests we conducted with the help of many volunteers, in the context of the Covid Community Alert project (see this link for Portuguese). Details about the methodology and the data we collected are described below.
Table of contents¶
* [Introduction](#introduction) * [Data Description (Metadata)](#data-description-metadata) * [Data Summary](#data-summary)Introduction¶
We designed a custom Android application for collecting data on the precision of BLE for different phone models. The app was delivered to several volunteers, who ran the tests on pairs of mobile phones that they had available.
Test Methodology¶
Each test was run on two devices at a time. The volunteers were asked to read and follow the instructions presented in the app, which consisted of:
- Positioning the devices X meters apart from each other, on a height close to the pocket of a person standing up or over a chair;
- Waiting for 2 minutes while logs of the advertisements via BLE were recorded. The devices were set to continuously advertise for 30 seconds and sleep for 5 seconds.
The aforementioned steps should be executed for X = 0.5, 1, 2, and 4 meters. All the tests were automated so that the volunteers only had to position the devices accordingly and press a button when ready.
Data Description (Metadata)¶
Files (available here):
- bluetooth_devices.csv
- bluetooth_friendly_models.csv
- bluetooth_tests.csv
- bluetooth_logs.csv
Devices (bluetooth_devices.csv)¶
Identification and description of the devices, including brand, manufacturer, model number, and Android version.
| beacon_id | bluetooth_le | advertising_supported | manufacturer | brand | model | incremental | version_code | base | create_date | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 684b4852-8f94-4c13-ba39-effd16914c82 | True | 0 | motorola | motorola | moto g(6) play | 3bd1c | 9 | 1 | 2020-04-10T12:57:37.618Z |
| 1 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | True | 0 | motorola | motorola | Moto G (5) | 3447de | 8.1.0 | 1 | 2020-04-10T12:59:47.150Z |
| 2 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | True | 0 | motorola | motorola | Moto G (5) | 3447de | 8.1.0 | 1 | 2020-04-10T12:59:48.783Z |
| 3 | 769bf0c9-62ed-4b9e-9dda-5d1c54e3cc8a | True | 5 | samsung | samsung | SM-G610M | G610MUBS7CTA1 | 8.1.0 | 1 | 2020-04-10T13:33:55.476Z |
| 4 | cd500b77-5cfa-4e5a-877c-70e329157175 | True | 0 | Xiaomi | Xiaomi | MI 9 | V11.0.5.0.QFAMIXM | 10 | 1 | 2020-04-10T16:39:14.667Z |
Friendly Models (bluetooth_friendly_models.csv)¶
Correspondence of model numbers with their friendly name.
| model | friendly_model | |
|---|---|---|
| 0 | 4034E | Alcatel Pixi 4 |
| 1 | 5044Y | Alcatel U5 |
| 2 | ANE-LX2 | P20 Lite |
| 3 | ASUS_Z017DC | ZenFone 3 |
| 4 | BKL-L09 | Honor View 10 |
Tests (bluetooth_tests.csv)¶
Records of when tests were started by the volunteer through the app. Each record identifies the device, the distance being tested, and the date and time when the volunteer started the test.
| beacon_id | real_distance | timestamp_start | create_date | |
|---|---|---|---|---|
| 0 | 684b4852-8f94-4c13-ba39-effd16914c82 | 0.5 | Apr 10, 2020 09:54:39 | 2020-04-10T12:57:58.799Z |
| 1 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | 0.5 | Apr 10, 2020 10:00:34 AM | 2020-04-10T13:00:35.746Z |
| 2 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | 0.5 | Apr 10, 2020 10:00:36 | 2020-04-10T13:00:37.697Z |
| 3 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | 1 | Apr 10, 2020 10:03:24 AM | 2020-04-10T13:03:25.733Z |
| 4 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | 1 | Apr 10, 2020 10:03:30 | 2020-04-10T13:03:31.494Z |
Logs (bluetooth_logs.csv)¶
Logs recorded for each device, including the RSSI, the real distance, and the distance estimated by the device (through the altbeacon library).
| beacon_id | receiver_id | real_distance | timestamp | create_date | measured_distance | rssi | |
|---|---|---|---|---|---|---|---|
| 0 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | 0.5 | Apr 10, 2020 10:00:35 AM | 2020-04-10T13:00:36.665Z | 0.648581 | -52 |
| 1 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | 0.5 | Apr 10, 2020 10:00:36 AM | 2020-04-10T13:00:37.928Z | 0.552367 | -52 |
| 2 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | 0.5 | Apr 10, 2020 10:00:37 AM | 2020-04-10T13:00:38.834Z | 0.466343 | -45 |
| 3 | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | 0.5 | Apr 10, 2020 10:00:38 AM | 2020-04-10T13:00:39.996Z | 0.301484 | -51 |
| 4 | b24147bc-55b6-41dc-9ffc-f90a8ff65fdb | ac2066a0-fcac-4d90-b25c-30d9a63f9ad9 | 0.5 | Apr 10, 2020 10:00:39 | 2020-04-10T13:00:40.234Z | 0.143828 | -47 |
Data Summary¶
Tests were completely run on 65 different devices, although logs were recorded for only 60 of them. An overview of the RSSI distribution and the distances estimated by the devices is presented in the following heatmaps, according to the tested distance:
The number of devices with logs recorded for each tested distance is shown below, grouped by manufacturer.
The recorded RSSI distributions are shown below for each manufacturer.
The distribution of the distances estimated by the library from the RSSI are presented below:
The RSSI data for each individual model is plotted below, considering only the three most frequent manufacturers in our dataset:
