Date of Award

8-2025

Document Type

Thesis

Degree Name

Master of Science

Degree Discipline

Electrical Engineering

Abstract

Accurate beam-level traffic forecasting is critical for 5G wireless network optimization, yet is challenged by inherent data sparsity and multi-scale temporal patterns. This thesis presents a novel Gated Recurrent Unit (GRU)-based Multi-Task Learning (MTL) framework with ensemble enhancement for spatio-temporal traffic prediction. The approach used shared representations to jointly perform classification and regression, making it more effective than traditional methods at handling beam-level traffic prediction. Through systematic evaluation of six machine learning models (Linear Regression, DLinear, XGBoost, Echo State Network, Long Short-Term Memory, and GRU-MTL) across three sequence lengths (168/24/8-hour), we demonstrated that: (1) the proposed MTL approach outperformed conventional machine learning approaches (for example, Mean Average Error (MAE)=0.3223 for LSTM vs. MAE=0.2136 for MTL using 168-hour sequences), leveraging shared representations to address intermittent beam activity; and (2) a weighted ensemble of top-performing models (MTL, XGBoost, Linear Regression) achieved further 1.45% improvement. Using AI for Good real-world data, we established that weekly horizons optimized accuracy for capacity planning, while the ensemble maintained robustness across temporal scales. These results offer a framework for sparse spatio-temporal traffic forecasting in 5G, balancing accuracy and practical deployment for resource allocation and congestion control.

Index Terms: 5G, DLinear, ensemble, echo state network (ESN), gate resource unity (GRU), long short-term memory (LSTM), multi-task learning, time series prediction, traffic forecasting, XGBoost.

Committee Chair/Advisor

Xiangfang Li

Committee Member

Annamalai Annamalai

Committee Member

Lijun Qian

Publisher

Prairie View A&M University

Date of Digitization

10/21/2025

Contributing Institution

J. B . Coleman Library

City of Publication

Prairie View

MIME Type

Application/PDF


Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.