Machine Learning Based Throughput Prediction for 4G/5G Mobile Networks Using Large Scale Dataset
| dc.contributor.author | Ali Jamoos | |
| dc.date.accessioned | 2026-09-06T11:13:40Z | |
| dc.date.available | 2026-09-06T11:13:40Z | |
| dc.date.issued | 2026-08-25 | |
| dc.description.abstract | As cellular networks evolve from 4G LTE to 5G, throughput prediction becomes increasingly important for understanding user experience and supporting data-driven network optimization. However, real active-measurement datasets are often sparse, noisy, and technology-dependent, making throughput modeling a challenging regression problem. This paper presents a machine-learning-based framework for predicting downlink and uplink throughput in both 4G and 5G networks using large-scale active measurement dataset. The original dataset contains 367,104 samples and 81 parameters, including radio indicators, scheduling features, modulation statistics, transport block size information, and contextual variables. Due to the high sparsity of the throughput targets, the data were divided into four independent regression tasks: 4 G downlink, 4 G uplink, 5 G downlink, and 5 G uplink. After preprocessing, feature ranking, and outlier removal, several regression models were trained and evaluated using 10-fold cross-validation in MAT-LAB Regression Learner. The results show that tuned Bagged Trees achieved the best performance across all four datasets, reaching R2 values of 0.9017, 0.9605, 0.8436, and 0.8767 for 4G DL, 4G UL, 5G DL, and 5G UL, respectively. These results demonstrate that target-specific ensemble learning can effectively model the nonlinear relationship between radio/scheduling parameters and measured cellular throughput. | |
| dc.identifier.citation | H. Bahar, T. Sider and A. Jamoos, "Machine Learning Based Throughput Prediction for 4G/5G Mobile Networks Using Large Scale Dataset," 2026 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies (AEECT), Amman, Jordan, 2026, pp. 479-484, doi: 10.1109/AEECT69724.2026.11657828. | |
| dc.identifier.other | https://ieeexplore.ieee.org/document/11657828 | |
| dc.identifier.uri | https://dspace.alquds.edu/handle/20.500.12213/10737 | |
| dc.language.iso | en | |
| dc.title | Machine Learning Based Throughput Prediction for 4G/5G Mobile Networks Using Large Scale Dataset | |
| dc.type | Article |