SliceVision256: A Multi-Paradigm Synthetic Dataset for Network Slice Visualization

Overview

SliceVision256 is a comprehensive synthetic dataset containing 12,000 high-resolution (256×256) images of network slice visualizations generated through four distinct algorithmic approaches. Designed for machine learning research in 5G/6G network slicing, computer vision, and pattern recognition, this dataset provides unique representations of three key network slice types:

• eMBB (Enhanced Mobile Broadband)
• URLLC (Ultra-Reliable Low-Latency Communications)
• mMTC (Massive Machine-Type Communications)

Each generation method captures different aspects of network behavior through specialized visualization paradigms.

Dataset Samples

Generation Method	Sample Visualization
Wave-Based Procedural
Fractal-Based
Agent-Based
Graph-Based

Dataset Structure

AbidHasanRafi/SliceVision256/
├── Agent-Based Synthetic Network Slice Images/
│   ├── metadata.csv
│   └── images/          # 3,000 agent-based slice images
├── Fractal-Based Synthetic Network Slice Images/
│   ├── metadata.csv
│   └── Images/          # 3,000 fractal-based slice images  
├── Graph-Based Synthetic Network Slice Images/
│   ├── metadata.csv
│   └── images/          # 3,000 graph-based slice images
└── Wave-Based Procedural Synthetic Network Slice Images/
    ├── metadata.csv
    └── images/          # 3,000 procedural slice images

Generation Methodologies

1. Wave-Based Procedural Generation

• Technique: Mathematical waveform synthesis (sinusoidal + Poisson patterns)
• Characteristics:
  • Red channel: Bandwidth as multi-frequency waveforms
  • Green channel: Latency as directional wavefronts
  • Blue channel: Reliability as Gaussian zones
• Class Specialization: Burst traffic (eMBB), priority paths (URLLC), periodic reporting (mMTC)

2. Fractal-Based Generation

• Technique: Perlin noise with multi-octave persistence
• Characteristics:
  • Organic, self-similar patterns
  • Channel-specific noise scaling
  • Class-specific modifications (bursts, directional emphasis, sparse points)

3. Agent-Based Simulation

• Technique: Autonomous agent modeling
• Characteristics:
  • eMBB: Random walk with traffic bursts
  • URLLC: Linear low-jitter paths
  • mMTC: Sparse device activation patterns
• Visual Output: Smoothed trajectory heatmaps

4. Graph-Based Generation

• Technique: Network topology rendering
• Graph Types:
  • eMBB: Scale-free (Barabási–Albert) networks
  • URLLC: Grid networks
  • mMTC: Random geometric graphs
• Rendering: Edge-based traffic simulation with node effects

Metadata Schema

All subdirectories contain a metadata.csv with:

image_path, slice_type, bandwidth, latency, reliability

Usage Terms

This dataset is part of ongoing research. All rights reserved by the authors. Any use of this dataset must comply with:

1. Academic Use: Reference required
2. Commercial Use: Prohibited without any consent
3. Redistribution: Not permitted without authorization

Contributors

• Algorithm Design: Md. Abid Hasan Rafi, Mst. Fatematuj Johora
• Validation: Mohima Binte Rasel
• Institutional Affiliation: Department of Electronics and Communication Engineering, HSTU, Dinajpur, Bangladesh