Ateme has been showing how Pay TV providers can reduce cost and carbon from unicast streaming with a solution that automatically increases compression on popular programming by taking CPU capacity away from channels that currently have small audiences – with the result that the lower-audience shows are compressed more lightly. Picture quality is not affected.
The lower-audience programme gets the same picture quality using less CPU resource, but at the expense of more bandwidth. The higher-audience show gets the same picture quality in less bandwidth by using more CPU resource. The total CPU requirement can be balanced so it does not increase.
The idea is that reducing the bitrate on a streamed programme that is unicast to hundreds of thousands of people, for example, results in large total bandwidth savings. This bandwidth reduction outweighs the increased bandwidth requirements for a show that is only unicast to tens of thousands, for example. The system responds to CDN data showing demand for the content. This juggling act – involving CPU capacity, the degree of compression, bandwidth savings and bandwidth increases – can be played out across the whole channel portfolio.
This capability builds on existing solutions from Ateme to reduce energy consumption without damaging QoE for viewers. For example, feedback between the CDN and encoders can be used to remove bitrate profiles from the ABR bitrate ladder if they are not being watched. The Ateme technology, which falls under the banner of ‘Audience-Aware Streaming’ (first announced in March 2023) will then look at the new bitrate ladder and decide if two remaining profiles are too similar, and remove any near-duplication. The result is, effectively, a new bitrate ladder with a balanced spectrum of bitrate profiles, all of which are being used.
Another piece of the end-to-end jigsaw proposed by Ateme, when looking to reduce energy consumption or cost, is the ‘elastic CDN’ concept. This is where a cloud-native CDN node can perform multiple functions and switch between them within minutes. So, if there are few viewing requests, the node capacity can be turned to something like file transcoding instead. Then, if viewing demand grows, capacity is given back to classic CDN stream delivery. Thus, the CDN scales up to meet peaks but avoids sitting idle between those peaks.
