Network-Canvas

Introduction

Network Canvas is a project built as a solution for Bus Division Cellular Coverage Study offered by GL Communications,Inc and University of Maryland for the Info Challenge 2025.

This project evaluates and compares the robustness and reliability of cellular communication links from three different service providers within an indoor bus maintenance facility.

Project Overview

Network Canvas transforms complex signal data into actionable insights through interactive visualizations, empowering transit agencies to make informed decisions about cellular coverage within their facilities.

Challenge Statement

A transit agency needed to evaluate cellular coverage across three carriers in their indoor bus maintenance facility. The objectives were to:

• Compare cellular carriers’ performance
• Verify adequate cellular data coverage
• Identify any major coverage gaps
• Provide recommendations for repeater placement (optional)

Methodology

The project utilized a combination of data collection, analysis, and visualization techniques to achieve the objectives. The methodology involved:

1. Data Preprocessing

• Applied a data science lifecycle approach to process three carrier datasets

• Identified and removed null values to ensure data integrity

• Discovered and resolved duplicate measurements within quads

• Implemented dual processing strategies:

  • Average values method for baseline comparison
  • Maximum values method for peak performance analysis

2. Exploratory Data Analysis

Developed interactive visualizations for multiple metrics:

• Signal strength (RSSI)
• TCP Upload Speed
• Quality of Service (QoS)
• Round Trip Time (RTT)

Created comprehensive statistical dashboards with:

• Histograms showing distribution of metrics
• Box plots revealing statistical ranges
• Violin plots highlighting carrier differences

Also, have a scatter plot to view the relationship between the 3 datasets.

3. Coverage Visualization

• Implemented interactive heatmap visualizations of cellular performance
• Color-coded display indicating performance categories from “Poor” to “Good”
• Grid-based representation of the facility showing metric performance by location
• Toggle functionality to switch between metrics and carriers

4. Repeater Placement Algorithm

• Designed an optimization algorithm to identify optimal repeater locations
• Modified standard coverage formulas to better fit the indoor facility context
• Visualized recommended repeater placement with interactive overlays
• Prioritized locations based on maximum coverage improvement potential

5. Comparative Analysis

• Developed normalized scoring system across all critical metrics
• Created dual-scoring methodology for both average and peak performance
• Generated weighted metrics based on importance for transit operations
• Visualized comparative performance through heatmaps and summary statistics

Results & Recommendations

• Identified the best-performing carrier based on weighted metrics
• Provided strategic repeater placement recommendations
• Delivered interactive dashboard for monitoring

How to run the code

Prerequisites:

• Git
• Python 3.6+

1. Clone the repository

git clone https://github.com/Eyepatch0/Network-Canvas

cd Network-Canvas

2. Create a virtual environment

python -m venv venv

# Activate it (Windows)
venv\Scripts\activate

# Activate it (Linux/macOS)
source venv/bin/activate

3. Install the required packages

pip install --upgrade pip
pip install -r requirements.txt

4. Run the code

# Option 1: Run with Jupyter Notebook
jupyter notebook

Note:Once Jupyter launches, select the correct kernel matching your virtual environment, then run the notebook cells.

Quick Visualization Mode

To view only the visualizations without running any code: bash

• Run Voila to view the notebook as an interactive web app.

voila cellular_coverage_study.ipynb

Then open http://localhost:8866 in your browser to view the interactive visualizations.

Future Work

• Integrate real-time data feeds for dynamic monitoring
• Expand coverage to additional facilities
• Incorporate machine learning for predictive analytics
• Enhance user interface for improved usability

AI Usage:

Throughout the Network Canvas project, I used AI to improve code efficiency where appropriate. After writing functional but repetitive code for the repeater optimization algorithm, I leveraged language models to refactor into cleaner loop-based solutions. This allowed me to maintain focus on the analytical methodology while ensuring code quality. AI served as a programming assistant, enhancing implementation without replacing my original problem-solving approach.

• Used Claude 3.7Sonnet to refactor code
• Used Perplexity (Deep Research) to save time researching contextual information

Technologies Used

Can be found in the requirements.txt file.

Future Work

• Integration with real-time cellular data collection
• Expansion to additional facilities
• Incorporation of machine learning for predictive analytics
• Enhanced user interface for improved usability
• Improved documentation and user guides
• Additional metrics and analysis features

Contact

You can reach out to me on LinkedIn or GitHub for any questions or feedback. I am always open to suggestions and improvements.

Acknowledgements

Special thanks to the University of Maryland and GL Communications, Inc. for providing the opportunity to work on this project. I would also like to thank Dr. Kam F. Yee, and my mentors for their support and guidance throughout the project.