Advanced Analytics for Video Streaming
Video streaming comprises over 50% of the traffic carried over the Internet, and hence it is vital that both the video service providers and network operators have the ability to analyze streamed or OTT video. Streamed video has the characteristic that the transmission process (over TCP) does not introduce visual artifacts however the streaming process may suffer from slow network connections or retransmission that results in rebuffering and short term video playback freezes.
The analysis of video streams has a number of challenges.
(i) Protocol implementations vary
Video services often use proprietary protocols and, even if they use standard protocols and video formats, implementations vary considerably. This requires the video analytics software to support a wide and growing range of video formats and protocols.
(ii) Video streams are encrypted
Increasingly, OTT video is streamed using HTTPS and hence is virtually indistinguishable from an encrypted data session. Video analytics software has to use behavioral analysis of the encrypted stream to detect that the video is being carried, if video is detected then it can continue to analyze the stream. As the Index or Manifest that incorporates video codec, frame interval and image size will also be encrypted then the analytics software will also have to make educated assumptions about these parameters.
Telchemy's VQmon is able to analyze Adobe HDS, Apple HLS, DASH, Microsoft Silverlight (or Smooth Streaming) and some proprietary video streaming protocols. It also uses behavioral analysis to detect encrypted video streams and to extract key information about the stream. VQmon produces video MOS scores that are computed using video stream format, the streaming process and rebuffering analysis and produces a wealth of metrics related to the video streaming process. VQmon is widely used within the industry for OTT video analytics, for example the performance analysis of video streamed over 4G mobile connections.
The analysis of video streams has a number of challenges.
(i) Protocol implementations vary
Video services often use proprietary protocols and, even if they use standard protocols and video formats, implementations vary considerably. This requires the video analytics software to support a wide and growing range of video formats and protocols.
(ii) Video streams are encrypted
Increasingly, OTT video is streamed using HTTPS and hence is virtually indistinguishable from an encrypted data session. Video analytics software has to use behavioral analysis of the encrypted stream to detect that the video is being carried, if video is detected then it can continue to analyze the stream. As the Index or Manifest that incorporates video codec, frame interval and image size will also be encrypted then the analytics software will also have to make educated assumptions about these parameters.
Telchemy's VQmon is able to analyze Adobe HDS, Apple HLS, DASH, Microsoft Silverlight (or Smooth Streaming) and some proprietary video streaming protocols. It also uses behavioral analysis to detect encrypted video streams and to extract key information about the stream. VQmon produces video MOS scores that are computed using video stream format, the streaming process and rebuffering analysis and produces a wealth of metrics related to the video streaming process. VQmon is widely used within the industry for OTT video analytics, for example the performance analysis of video streamed over 4G mobile connections.