DaI: Decrypt and Infer the Quality of Real-Time Video Streaming

TL;DR

DaI is a novel traffic-based estimator that partially decrypts encrypted real-time video streams and uses machine learning to accurately assess quality of experience metrics, aiding network optimization.

Contribution

This paper introduces DaI, a method that combines partial decryption and machine learning to estimate video quality metrics in encrypted streams, addressing a key challenge in network monitoring.

Findings

Achieves 79% average accuracy in estimating QoE metrics.

Effectively handles encrypted real-time video streams.

Provides a practical tool for network quality assessment.

Abstract

Inferring the quality of network services is the vital basis of optimization for network operators. However, prevailing real-time video streaming applications adopt encryption for security, leaving it a problem to extract Quality of Service (QoS) indicators of real-time video. In this paper, we propose DaI, a traffic-based real-time video quality estimator. DaI can partially decrypt the encrypted real-time video data and applies machine learning methods to estimate key objective Quality of Experience (QoE) metrics of real-time video. According to the experimental results, DaI can estimate objective QoE metrics with an average accuracy of 79%.