# Characterizing the VPN Ecosystem in the Wild

**Authors:** Aniss Maghsoudlou, Lukas Vermeulen, Ingmar Poese, Oliver Gasser  
**arXiv:** [2302.06566](https://arxiv.org/abs/2302.06566)  
**Date:** 2023-02-13  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2302.06566v1/

## Key points

- COVID-19 pandemic has led to an increase in VPN usage globally
- Measuring traffic and security aspects of VPN ecosystem is now more important than ever
- Detecting and characterizing VPN traffic remains challenging due to some protocols using the same port number as web traffic
- Paper aims at detecting and characterizing VPN servers in the wild to facilitate identifying VPN traffic
- 9.8 million VPN servers distributed globally using OpenVPN, SSTP, PPTP, and IPsec protocols were identified through Internet-wide active measurements
- SSTP protocol is the most vulnerable among those detected with over 90% of servers being vulnerable to TLS downgrade attacks
- 2% of all servers that respond to their VPN probes also respond to HTTP probes and are classified as Web servers.
- The list of VPN servers was used to identify VPN traffic in a large European ISP's network, where 2.6% of all traffic was related to these VPN servers.
- Research provides valuable insights into the state of security for VPN solutions used by individuals and organizations worldwide.

## Summary

The COVID-19 pandemic has led to a significant increase in the use of Virtual Private Networks (VPNs) around the world, as remote work and online activities have become more prevalent. As a result, it is now more important than ever to measure the traffic and security aspects of the VPN ecosystem. However, detecting and characterizing VPN traffic remains challenging due to some VPN protocols using the same port number as web traffic. This makes port-based traffic classification ineffective, and VPN users are also concerned about their connections' vulnerabilities due to privacy issues. To address these challenges, this paper aims at detecting and characterizing VPN servers in the wild, which facilitates identifying VPN traffic. The authors perform Internet-wide active measurements to find VPN servers worldwide and analyze their cryptographic certificates, vulnerabilities, locations, and fingerprints. They identify 9.8 million VPN servers distributed globally using OpenVPN, SSTP, PPTP, and IPsec protocols. The study finds that SSTP is the most vulnerable protocol among those detected with over 90% of servers being vulnerable to TLS downgrade attacks. Additionally, out of all the servers that respond to their VPN probes, 2% also respond to HTTP probes and are classified as Web servers. The authors apply their list of VPN servers to identify VPN traffic in a large European ISP's network and observe that 2.6% of all traffic is related to these VPN servers. This research provides valuable insights into the state of security for VPN solutions used by individuals and organizations worldwide. The findings can help improve security measures for both users and providers while highlighting areas where further research is needed to enhance privacy protection for online activities.