A Method For Identifying Resolutions And Bitrates Of Youtube Encrypted Video Traffic In Mobile Network

With the continuous development of mobile networks,the number of mobile video viewers is also increasing.Massive mobile traffic data presents an enormous challenge to mobile network service providers who want to ensure that video streams will not affect the online experience of other users,and mobile video viewers have a superior video viewing experience.Therefore,mobile network service providers need to analyze the popular video traffic and recognize the bitrates and resolutions of videos that users watched.But with the video streams encrypted by HTTPS,how to obtain video bitrates and resolutions from HTTPS encrypted traffic has become a hot topic in the field of traffic identification.In this paper,we propose a scheme to identify the bitrates and resolutions of YouTube encrypted traffic in mobile network.The main contents and contributions of this paper are as follows:YouTube data chunks recognition:YouTube encrypted video traffic contains not only encrypted traffic characteristics but also multimedia traffic characteristics.At present,research towards encrypted traffic is mostly based on data flow analysis.This method is simple and clear and can quickly analyze the characteristics of encrypted traffic.However,this analysis result is not good enough for YouTube encrypted video traffic.This paper analyzes the segmented transmission characteristics of YouTube encrypted video traffic,and proposes a YouTube data chunk recognition method based on ACK Number.We segment encrypted traffic into chunks according to the ACK Number message and identify the data chunks YouTube server transferred.The experimental results show that this method can accurately identify YouTube data chunks from the encrypted traffic.YouTube video bitrates recognition:YouTube video bitrates recognition is divided into Android platform bitrates recognition and IOS platform bitrates recognition.Since the audio source and video source of YouTube videos are transferred in separate,it is necessary to identify audio chunks and video chunks from data chunks after YouTube data chunk recognition.At the same time we need to deal with the fragments caused by network interruption:end point continued for Android,retransmission for IOS.As video viewers are in the adaptive resolution mode of play,this paper removes data chunks that had been cached but discarded by video player and calculate the video chunks' playback time.In the end this paper calculated the bitrates of the video chunks.The experimental results show that this method can accurately identify the bitrates of each video chunk of the video watched.YouTube video resolution recognition:As YouTube applies the video segmentation mechanism to both the Android platform and the IOS platform,recognizing the resolution of video chunks in encrypted traffic becomes the key to YouTube video resolution recognition.This paper presents a method of resolution recognition based on the characteristics of video chunks.This method extracts five attribute features from video chunks,and then uses C4.5 decision tree algorithm to analyze and generate resolution recognition rules.Finally,k-means algorithm is introduced for recognition result correction.Experimental results show that this method can more accurately identify YouTube video resolution information.