What is the alternative to unicast? To handle the 10’s of millions of customers expected to start consuming video across Ip . when they begin shifting the technology from Cable/Sat to IP? ( needs solving )
Imagine an end pint service that an ABR client can query in order to know what is the current health of the network IS. In Order to select which Bitrate it should play? Its Content Aware streaming but Something  an open spec where anyone can access the data. A lot of vendors do this but imagine something which was open and useable by clients. Something like a NTP time service which you can Query and get a response back from the edge node.
In addition to buffering we should start thinking how we can measure the Perceptual video of the playback experience. Instead of looking at the average bitrate of a session we should start looking at the perceptual score on average ( something like VMAF = 85 on average per a video play session. The key would be do design a spec / method on how to create the Scores at source, then collect and measure.
implications and recommendations for streaming
implications and recommendations for streaming
elevate live streaming latency work
how about a best practices or guidelines for UI / UX for a streaming product?
As more and more content becomes available online it becomes increasingly more difficult for viewers to find the content they want to consume and publishers to target and engage with their audience.
How the infrastructure supply chain is going to need to change in a world where UHD and live streaming are becoming more prevalent.
Tuning the live content workflow and delivery so that IP delivery is faster than broadcast or satellite delivery.
develop plugins for major caching/reverse-proxy providers like NGINX (now F5), Squid, Varnish, etc.
an industry standard for metadata elements and representation to describe video content assets; build off MovieLabs Ontological framework and EIDR
develop a personal ID for people (like IMEI) that can be used across OTT services to track what a user watches in order to be leveraged for content recommendations. SVA becomes “clearing house” and storage for the dataset (as a neutral, non-commercial third-party in order to avoid VPPA issues).
No specific definition provided.
the DASH gap, HLS has a mechanism for I-frame index, but DASH has no specific mechanism. The intent would be to promote a standard in-band method for supporting content trick tracks/scrubbing
hand off from content producers to service/streaming providers, etc.
The solutions are proprietary and vary in capabilities. I think it would be good to look at whether some form of standardization would be of benefit. I view this as a critical tool for such use cases (further illustrated by DAZN) and capital avoidance for ephemeral events. It would be great if we could extend to other events such as patches, game updates, etc. that drive significant utilization for short periods of time (although this is a bit outside the SVA's purview).
As we continue to address the various causes of buffering within our control, we will eventually hit diminishing returns due to last mile challenges. Wireless networks (WiFi in particular) bring challenges for OTT streaming that we may be able to drive some work in. Possibly partnering with the WiFi industry to categorize the types of WiFi impairments and the associated response/remediation and its impact on OTT delivery. WiFi being a broadcast medium brings some unique challenges that wired point to point technologies don't have to worry about. I think the 5G folks have kept this in mind as they design their new technology but I am not sure the WiFi industry is thinking about this.
this is the idea of localizing (likely a pre-localization step with the backend CMS) client requests based on the content being asked for. This allows us to direct the client to the cache in the closest that is responsible for the content. The goal is to make the edge more efficient and have less 'serve to peer' traffic which drives up utilization and performance (by reducing hop count).