Internal Optimisation of Network Streaming
Project Status: cancelled
Start Date: April 2020
End Date: March 2022
Budget (total): 2386 K€
Effort: 18 PY
Project-ID: C2019/2-3
BT, United Kingdom
Genesis Technology Services, United Kingdom
Limelight Networks (UK), United Kingdom
Enensys, France
IRT, Germany
Abstract
This project will use AI techniques to allow 5G networks to employ multi cast capability in a dynamic way as an internal optimisation tool. The approach is also expected to be applicable to other networks.
Content types such as such as broadcast TV, live events, computer games and updates for mobile apps and operating system software drive infrequent but significant traffic peaks caused by near synchronised downloading of content. Network investment decisions are driven by the need to accommodate these infrequent peaks. The tendency of synchronised content downloads to drive network investment will increase as public service broadcasters deliver more of their content over the Internet.
The goal of this project is to use an AI to support an innovation in that will make 5G (and other networks) better able to deliver cost-effective, high-quality live content experiences. It will do this by pursuing the development of a meaningful 5G broadcast capability attempting to instantiate broadcast prototypes based on proposals from earlier projects. Secondly, the project will attempt to develop a dynamic, AI driven system which satisfies the needs of the key players in the content delivery chain.
A central principle of our approach will be to treat multicast as an internal network optimisation tool, rather than a service to be sold in its own right. This principle distinguishes our approach from all previous hybrid unicast/multicast delivery activities and makes the investment case for the proposed multicast infrastructure easier. It allows network operators to upgrade networks regionally, rather than depending on national roll out, and allows them to build a business case on internal cost savings rather than on the uncertain income from future customers of an end-to-end multicast network. It also means that not all devices have to support multicast to make it a viable efficiency measure.