A multi-device and multi-operator dataset from mobile network coverage on Android devices

The demand for mobile coverage with adequate signal quality has triggered criticism due to the maturity of the Internet's diffusion in today's society. However, with the deployment of 5G networks, even 5G NSA by 4G LTE, the complexity of the operating environment of mobile networks has increased. To evaluate the behavior of mobile networks in terms of signal quality and other important metrics for mobile telephony, we developed a dataset consisting of 33 radio parameters that can collect up to 736,974 records generated daily by smartphones and tablets. To create the dataset, an application was designed for the Android operating system using the Kotlin programming language, which can collect data in real time and generate a CSV file.
The dataset has 10 samples collected from 9 cities located on the Amazon and Negro Rivers. The complete database covering all regions has 33 columns and 736,974 rows. In addition to the primary dataset, we divided the data into three regions: the metropolitan area of Manaus, the middle Solimões River, and the middle Amazonas River. During the scheduled trips, data were collected along rivers and roads that provide access to the locations selected for the experiment. The data was processed, indexed, and organized into a comprehensive database, then categorized by location. This organization allows experiments using the entire dataset across all cities or with data specific to an individual city. To access the database and conduct initial experiments, Python scripts were developed alongside the database to facilitate data loading and the generation of histograms and charts necessary for the initial investigation. In addition to the graph generation scripts, we also created heat maps based on the collected network variables.The data is organized in a folder named “network_dataset,” which contains a list of datasets. Each dataset is named according to the device ID concatenated with the timestamp at which it was collected.The raw dataset was stored inside the mobile device, and stored in the cloud after the preprocessing steps. The collected data contains mobile network variables such as Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Signal-to-Noise Ratio (SNR) and Channel Quality Indicator (CQI), collected in real-time and stored on the mobile device in Comma-separated Values ​​(CSV) data format. After completing the daily collection, the device automatically sends the file to the cloud.