Beam-Level Traffic Timeseries
A massive dataset of time series representing 5G network traffic, useful for modeling, predicting, and optimizing the mobile network.
Description
Beam-Level Traffic Timeseries Dataset is a dataset structured in CSV files that contains detailed measurements of 5G network traffic per beam. It covers 2,880 beams spread across 30 base stations, with data such as downlink traffic volume, transmission time, PRB resource usage, and the number of users per beam. The data is organized into time series over several weeks, including segments for training and forecasting.
What is this dataset for?
- Network load forecasting to optimize 5G resource management
- Formation of time series models for mobile traffic
- Research on the modeling of user behaviors at the level of a cellular network
Can it be enriched or improved?
Yes. It is possible to add other data sources (weather, topography, etc.) to contextualize the measurements. Feature engineering techniques can also be applied to enrich time series. The dataset could be completed with uplink data, or annotations of network events (failures, congestions).
🔎 In summary
🧠 Recommended for
- Network data scientists
- Telecom engineers
- Time series researchers
🔧 Compatible tools
- Prophet
- PyTorch Forecasting
- TensorFlow
- Scikit-learn
- Pandas
💡 Tip
Combine this dataset with event calendars or usage data to improve your predictive models.
Frequently Asked Questions
Does this dataset make it possible to model network congestion?
Yes, PRB volume and usage data can be analyzed to detect or predict saturation situations.
Can it be used to compare different base stations?
Yes, the dataset contains data separated by station, cell, and beam, allowing for accurate comparative analyses.
Is it possible to apply machine learning directly to these files?
Yes, CSV files can be used directly with classic time series models such as LSTM or XGBoost.
