Cellular Traffic Forecasting and Analysis with Efficiency Transformer
- Publication Type:
- Thesis
- Issue Date:
- 2023
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
Cellular networks have witnessed an exponential growth in data traffic due to the proliferation of mobile devices and the increasing demand for high-bandwidth applications. Efficiently managing this burgeoning traffic has become a critical challenge for telecommunication providers. Forecasting and analyzing cellular traffic patterns are crucial for optimizing network performance, resource allocation, and ensuring a seamless user experience. In recent years, deep learning techniques have emerged as powerful tools for handling the complexity and dynamic nature of cellular traffic data.
This research presents a comprehensive review of statistical methods, machine learning-based methodologies and deep learning-based methodologies applied to cellular traffic forecasting and analysis. It provides an overview of traditional forecasting techniques and highlights the limitations that have prompted the adoption of deep learning models. Various deep learning architectures such as recurrent neural networks (RNNs), long short-term memory networks (LSTMs), convolutional neural networks (CNNs), and hybrid models are examined in the context of their applications for traffic prediction and analysis. Furthermore, this research discusses the challenges and opportunities associated with employing deep learning in cellular traffic management. Issues such as data heterogeneity, scalability, interpretability, and real-time processing constraints are addressed, along with potential solutions and future research directions.
The study discusses challenges and opportunities in employing deep learning for cellular traffic management, addressing issues like data heterogeneity and real-time processing constraints. Accurately forecasting mobile traffic data is a challenge due to the complex spatial and temporal correlations, especially when the mobile data comes from a large geographical area. To tackle this challenge, we propose a new model, called “ST-InducedTrans”, to dynamically explore the large geographical correlations (spatial) and periodic variations (temporal). Specifically, a Spatial Bottleneck Transformer is devised to obtain spatial correlations from the most relevant grids in the geographical area at the cost of linear complexity. Additionally, cross-domain datasets, including base station locations and social activity information, are incorporated to enhance prediction accuracy. Comprehensive experiments on real-world cellular data from Milan demonstrate the model's superiority over state-of-the-art methods in terms of accuracy metrics (MAE, NRMSE, and R2), with lower time complexity.
Overall, this research consolidates the current state-of-the-art in utilizing Transformer for cellular traffic forecasting and analysis, highlighting its potential to revolutionize how telecommunication networks anticipate and manage traffic demands, ultimately enhancing network efficiency and user satisfaction.
Please use this identifier to cite or link to this item: