Abstract:The fifth-generation (5G) mobile network has significantly enhanced people's lives with faster internet speed and more reliable connections. However, there is still insufficient coverage of 5G networks worldwide, requiring telecom operators to deploy more base stations to meet the increasing demand for 5G's further commercialization. In this regard, a major challenge is understanding user network behaviors and traffic demands in target areas where 5G has not yet been deployed, which is crucial for developing a more efficient base station deployment strategy. Mobile traffic generation is a potential approach that enables operators to preemptively estimate user network demands in target areas, thereby specifying corresponding deployment strategies to enhance network performance. However, existing methods have limitations in capturing spatio-temporal features of 5G mobile traffic, particularly in areas with insufficient 5G coverage and limited historical 5G traffic data. To fill this gap, we introduce a regional feature conditioned diffusion framework for 5G network traffic generation. Our models explore the relationship between 5G traffic and existing 4G traffic, utilizing a customized cross attention mechanism and graph convolutional networks (GCN) to capture the correlation between network traffic and regional features. Based on this relationship, the framework can characterize mobile network traffic demands, thereby achieving high-fidelity 5G traffic generation in target regions with insufficient 5G coverage. Extensive experiments on real-world datasets have shown that the proposed scheme outperforms state-of-the-art baselines by more than 10%, demonstrating its high-fidelity generation capability, controllability, and generalizability. Moreover, we have deployed our scheme on China Mobile's Jiutian Platform as a network traffic simulator to improve 5G base station deployment strategies.

DPCA: A Novel Strategy to Refine the Traffic Classification Results in 5G Context

FedEdge AI-TC: A Semi-supervised Traffic Classification Method based on Trusted Federated Deep Learning for Mobile Edge Computing

Novel Dynamic Multiple Classification System for Network Traffic

Network traffic grant classification based on 1DCNN-TCN-GRU hybrid model

THREE-DIMENSIONAL CONVOLUTIONAL NEURAL NETWORK BASED TRAFFIC CLASSIFICATION FOR WIRELESS COMMUNICATIONS

Capsule Network Assisted IoT Traffic Classification Mechanism for Smart Cities

Towards a Profiling View for Unsupervised Traffic Classification by Exploring the Statistic Features and Link Patterns

A Multi-Modal Approach For Context-Aware Network Traffic Classification

Research on Network Traffic Classification Based on Machine Learning and Deep Learning

Regional Features Conditioned Diffusion Models for 5G Network Traffic Generation

Traffic Flow Decomposition and Prediction Based on Robust Principal Component Analysis

A Model of Encrypted Network Traffic Classification that Trades Off Accuracy and Efficiency

Real-time Traffic Classification for 5G NSA Encrypted Data Flows With Physical Channel Records

Optimizing Feature Selection for Efficient Encrypted Traffic Classification: A Systematic Approach

A novel weighted combination technique for traffic classification

An Efficient Traffic Classification Scheme Using Embedded Feature Selection and LightGBM

Composite Lightweight Traffic Classification System for Network Management

A filter-and-refine approach to lightweight application traffic classification

Traffic Classification: Issues and Challenges

XAI Meets Mobile Traffic Classification: Understanding and Improving Multimodal Deep Learning Architectures