Identifying VoIP traffic in VPN tunnel via Flow Spatio-Temporal Features.

Area of Science:

• Computer Science
• Network Security
• Data Analysis

Background:

• Encrypted network communication and new applications complicate traffic analysis for network management and cybersecurity.
• Virtual Private Networks (VPNs) are widely used for bypassing censorship and accessing geo-restricted services, increasing the need to analyze traffic within them.

Purpose of the Study:

• To propose a novel identification scheme for Voice over Internet Protocol (VoIP) traffic specifically when it is tunneled through VPNs.
• To evaluate the effectiveness of this scheme in distinguishing VoIP flows from non-VoIP flows within VPN traffic.

Main Methods:

• Utilized a set of Flow Spatio-Temporal Features (FSTF) for traffic analysis.
• Employed and compared six distinct classifiers: decision trees, K-Nearest Neighbor (KNN), Bagging, Boosting via C4.5, and Multi-Layer Perceptron (MLP).

Main Results:

• The proposed scheme demonstrated high effectiveness in identifying VoIP traffic within VPNs.
• Performance was validated using metrics including overall accuracy, precision, sensitivity, and F-measure across the tested classifiers.

Conclusions:

• The developed scheme successfully distinguishes VoIP flows from non-VoIP flows within VPN traffic.
• This contributes to improved network management and cybersecurity by enabling better analysis of encrypted communications.