Key Technologies for the Content Distribution Framework

T4.2. Content Distribution Framework (M4 – M15) Task Leader: BT Participants: BBC, BLB, BPK, BT, EBU, EXP, LU, NOK, NOM, O2M, TUAS, UNIS The output of this task will be a high-level description of a framework for content distribution, which allows the combination of various optimisation techniques in an autonomous manner and will maintain a very simple interface with the content service provider, or global CDN operator. The following key technologies will be considered as part of this framework. - Multicast Operation On Demand MooD-like paradigms over various network types in support of our converged network vision will be a key optimisation tool. HTTP and derivatives, such as HTTP2 and Quick UDP Internet Connections, have become ubiquitous for non-conversational applications. HTTP is even the most common way of streaming video, despite being technically ill-suited to this task. The primary goal will therefore be the efficient delivery of HTTP (and derivative) traffic at scale using an optimum mix of unicast and multicast. A key aspect of this will be finding a suitable encapsulation format for the multicast content. Various proposals have been made to date (FLUTE, ROUTE, NORM), though none appears to be entirely suitable. This task will use the same (or very similar) encapsulation formats over all network types. It will combine MooD-like capability using CDN technology in the fixed network and MooD in the mobile network into a unified approach and drive towards converged autonomous MooD. In doing so, we will coordinate with task T4.2 of WP4. The respective benefits of using LTE Broadcast or CDN for medium popularity content will be assessed during the project when the use cases are refined (requirement for fixed/mobile convergence, availability of LTE Broadcast and CDN on base stations, etc.). This work will allow establishing the mechanism for selecting the best streaming methods (and allocating the streaming resources accordingly) and define the popularity thresholds for selecting the delivery method. - Information Centric Networking Information Centric Networking is used as a generic term for a number of techniques for content identification, discovery, caching and replication which are based on the concept of assigning unique names to pieces of content. Specific approaches, such as Content Centric Networking (CCN), represent a massive departure from a traditional caching model. Information Centric Networking (ICN) natively supports mobility, multi-access, unicast, multicast and broadcast. Although ICN is a popular research topic, there are challenges to be addressed before the industry accepts this new paradigm as the future of networking for relevant use cases. Not all of these challenges are purely technical. For example, CCN implies a rather different flow of money between infrastructure provider and content services provider which makes CCN appear more like a broadband ISP business, rather than an overlay CDN business. - Content Distribution Network In contrast to research approach of ICN, the traditional approach is of a well-planned CDN overlay in the network, consisting of a web of caches and smart request routing. However, the traditional approaches have to be significantly redesigned to work well with adaptive streaming and broadcast services for 5G-Xcast network. Specific challenges include: - Deep caching in the network, in conjunction with virtualization and edge cloud concepts for 5G. - Caching, and possibly predictive caching for segmented content (large number of small media files), network congestion and related cost issues. - Cache efficiency for adaptive streaming content (with multiple qualities). - Preventing quality oscillation issues occurring as a result of decentralized application logic interacting with caches. Probe application of server and network assistance, e.g. on the lines of MPEG (Moving Picture Experts Group), DASH SAND; here the term “network” refers to functional entitie