HBO Uses AI to Combat Buffering

Streaming Media

Conviva says quality internet streams are only possible with intelligent realtime detection, and HBO is using Conviva's new Video AI Platform to do just that with HBO GO and HBO NOW.

http://www.streamingmedia.com/Articles/News/Online-Video-News/HBO-Uses-AI-to-Combat-Buffering-with-Convivas-Help-120449.aspx#

Artificial intelligence (AI) is set to be a hot topic at IBC, with AI algorithms making the first meaningful contributions in media around real-time analysis and early warning of video delivery problems.

One company tackling the challenge is Conviva, which goes so far as to say that there's no getting away from the fact that to deliver TV over the internet you need artificial intelligence.

It is introducing an alerts program into its Video AI Platform to automatically detect and diagnoses service delivery issues. What's more, it has HBO on board as a reference customer.

"Publishers currently spend a considerable amount of time and effort to configure and monitor manual alerts to detect service delivery issues," explains Dr. Hui Zhang, co-founder & CEO of Conviva in a release. "This often requires large operational teams, who most times are not able to find nor configure every possible alert. Furthermore, these teams must spend valuable hours diagnosing the root cause of each and every issue. From our work with the world's largest publishers over the past ten years, we recognized the need for a solution addressing these pain points, leading to the development of Video AI Alerts."

The premise of AI as an efficient tool for OTT is acknowledgement that the internet isn't fit for purpose when it comes to video.

"Unfortunately, too many things can go wrong between the video server and the consumer's screen," Conviva states "Only AI can sort through the millions of permutations of possible problems to discover the ones that matter most as well as what will have the most impact on improving viewing experience."

Conviva's tool is underwritten by the Video AI Platform's detection and diagnosis artificial intelligence models. These models use analysis of billions of video streams reported by more than 2.5 billion sensors embedded within video players across Conviva's publisher network. According to the firm, this graph captures the relationships between entities within a publisher's catalog of content and the infrastructure used to deliver it – "much like the social graph that captures connections between various people and their interests. This graph is what enables the intelligence behind Video AI Alert's real-time diagnosis of root causes".

Hui Zhang says Video AI Alerts continually compares quality of experience (QoE) and engagement metrics against recent norms, and instantly detects anomalies.

"These alerts notify publishers of the issue, root cause, and likely solution thereby speeding resolution time," he says. "This eliminates the need to set filters and configure alerts based on static threshold values, reducing video stream disruptions and viewer dissatisfaction – this is not possible with manual alerts."

HBO's site reliability lead Vikrant Kelkar is on hand to explain that in one case, a stream "had been misconfigured and without the full stream URL reporting with each alert, we would not have been able to see nor diagnose the cause of this issue."

HBO is now using the AI alerts for streams of HBO GO and HBO NOW.

Artificial intelligence will contribute as much as $15.7 trillion to the world economy by 2030, according to a recent report by PwC. $6.6 trillion of this would come from increased productivity as businesses automate processes and augment their labour forces with new AI technology.

However, much of what is suddenly branded as AI in media is in fact the evolution of the data analysis solution market.  For example, AI can play an important role in improving accuracy and reliability for placement, it can combating redistribution piracy with proactive search and identification of illegal re-broadcasts and machine-learned predictive algorithms can use feedback loops to improve the relevance of the prediction engine over time.

Another important area is in network management. Bandwidth is not limitless and is a significant portion of the expense of operating, especially for OTT.

"Today's CDNs and protocols are very wasteful of bandwidth (your device will always pick the highest available bandwidth whether it's needed or not at that particular time," says Tim Child, CCO and co-founder at MAM vendor Cantemo. "Analyzing the video as it is transmitted and then using the data to determine the required bandwidth can help network operators make more efficient use of bandwidth, lowering costs and improving quality at the same time."