Estimating End-User Throughput Using Service Provider Cell Traces Via Gradient Boosting
Investigators
Team
- Ritika Bhatia
Ericsson partners
Research project
The adoption of 5G networks has enabled supporting applications that require high bandwidths and low latencies. Service providers need to manage their resources efficiently to avoid service interruptions. Therefore, being aware of customer’s experienced bandwidth is of paramount importance. The thesis focuses on accurately imputing missing values in the collected cell traces and consequently building a Regularized Gradient Boosting model to predict the user’s throughput using cell traces that are exclusively collected from the service provider’s resources. The objective is to predict the target variable (throughput) using cell traces from the 5G node. The pseudo-actual throughput values are mapped from the user equipment traces to the cell traces, and by using the reference of pseudo-actual throughput values, the throughput values are imputed in all the cell traces. The approach shows that we can accurately estimate the user equipment’s throughput by using the imputing and prediction approaches.
To enhance the scalability and reliability of the prediction models, it is necessary to train these models on a greater number of cell traces and user equipment traces that are captured in the different radio settings, such as connecting multiple user equipment to the same and different 5G cells, and various radio frequency conditions such as mid-band and high-band. Furthermore, the focus should be on collecting high-quality data in different radio settings.
Bhatia, R. (2022). Estimating End-User Throughput Using Service Provider Cell Traces Via Gradient Boosting. https://doi.org/10.22215/etd/2022-15100