Global Network Latency Analysis (RTT vs. Distance)

Overview

This project explores the empirical relationship between geographic distance and network latency (Round-Trip Time). Using a set of global Debian FTP mirror servers, I conducted active measurements from Milan, Italy, to verify the linear correlation between distance and delay.

Theoretical Model

The analysis is grounded in the theoretical model for RTT: $$RTT = 2(L/C + d/v) + n$$ Where:

• $d$ is the physical distance.
• $v$ is the propagation speed.
• $n$ represents fixed processing delays at network elements.

Methodology (BYOD Approach)

Following the Bring Your Own Device (BYOD) principle from the Network Measurement Lab at Politecnico di Milano, I performed local measurements to ensure geographic accuracy:

1. Source Point: Milan, Italy (Local Machine).
2. Tools: A custom Python script using ping3 for latency and geopy for geodesic distance calculation.
3. Target Endpoints: Diverse servers across Italy, Europe, USA, Brazil, Japan, and Australia.

Key Results

• Estimated RTT increase per kilometer: 0.01800 ms/km.
• Calculated Propagation Speed: ~111,111 km/s (reflecting real-world routing overhead compared to the speed of light in fiber).

Project Structure

• my_measurements.py: The local data collection script.
• Network_Measurement_HW1.ipynb: The Jupyter Notebook used for data visualization and linear fitting.
• LICENSE: MIT License.

Course Context

This project was developed for the Network Measurement and Data Analysis Lab course, taught by Prof. Alessandro Redondi and Prof. Francesco Musumeci.