Large-Scale 802.11 Wireless Networks Data Analysis Based on Graph Clustering.

Area of Science:

• Computer Science
• Network Engineering
• Data Science

Background:

• Real-world Wi-Fi networks in over 1,200 educational centers were analyzed.
• Data from 9,000+ access points (APs) over one year provided insights into school Wi-Fi usage.
• RSSI measurements were used to construct Wi-Fi network conflict graphs for each school.

Purpose of the Study:

• To develop a method for analyzing Wi-Fi network behavior in educational settings.
• To assess the utility of graph embeddings for understanding wireless network dynamics.
• To identify network variability and enable configuration reuse across similar schools.

Main Methods:

• A large-scale dataset of Wi-Fi operating networks was collected and analyzed.
• A novel graph embedding technique based on classical graph features was proposed for Wi-Fi conflict graphs.
• Clustering techniques were applied to analyze temporal variations in conflict graphs.

Main Results:

• The proposed embedding demonstrated high discrimination power among different school Wi-Fi networks.
• The methodology successfully separated stable network scenarios from those with high variability.
• Network behavior patterns were compared across schools, enabling the identification of reusable optimal configurations.

Conclusions:

• Graph embeddings offer a powerful tool for analyzing and understanding complex Wi-Fi network dynamics.
• The developed approach aids in resource allocation by identifying networks requiring optimization.
• This study facilitates the transfer of successful network configurations across similar educational institutions.