E-Ensemble: A Novel Ensemble Classifier for Encrypted Video Identification

In recent years, video identification within encrypted network traffic has gained popularity for many reasons. For example, a government may want to track what content is being watched by its citizens, or businesses block certain productivity. Many such reasons advocate the need users on internet. However, with introduction of secure socket layer (SSL) and transport security (TLS), it become difficult analyze traffic. addition, dynamic adaptive streaming over HTTP (DASH), which creates abnormalities due variable-bitrate (VBR) encoding, makes researchers identify videos in internet The default quality settings browsers automatically adjust depending load. These auto-quality also increase challenge detection. This paper presents novel ensemble classifier, E-Ensemble, overcomes To achieve this, three different classifiers are combined using two combinations classifiers: hard-level soft-level combinations. verify performance proposed were trained dataset collected one month tested separate captured 20 days at date time. combination showed more stable results handling than those combination. Furthermore, classifier technique outperformed high accuracy 81.81%, even mode.