A Network based Method for Quality Estimation of TCP based Progressive Download Video Services

Inhalt:
Video streaming services, such as YouTube, podcasts or ZDF Mediathek services are responsible for a major part of the consumed data volume in the Internet and it is expected, that they will also strongly affect mobile networks. User satisfaction and user experienced service quality are strongly correlated and lead - from an Internet service provider point of view - either to an increase in subscription numbers or to customer churn towards competitors. For that reason, the streaming video quality needs to be permanently monitored. In our approach we propose an advanced network based method for real-time progressive downloaded video quality estimation. The biggest advantage of this network based quality monitoring method is the client’s end device and server platform independent way of operation. The proposed Quality Monitoring (QMON) component has its measurement point within the operators’ network. QMON performs a play-out buffer fill-level estimation based on the measured TCP segments and ACKs of the video stream passing through the observation point. Besides the basic offline and online estimation method three differently powerful flavours of the estimation algorithm are presented in the paper. The QMON method has already been implemented and covers a wide range of video codecs like the high compressed h.264 codec which is mainly used in the MP4 container as well as the new VP8 codec provided by Google in WebM container based download services. Also the former commonly used Flash Video service type is supported. The cutting edge scalable video codecs like Dynamic Adaptive Streaming over HTTP (DASH) also known as MPEG-DASH or Scalable Video Codec (SVC) are also targeted with the QMON method and are considered in this paper within a conceptual study.