Technical Blog
IPTV Instant Expert Part 2: The CDN (Content Distribution Network)
7th November 2011
The CDN (Content Distribution Network) is an end to end infrastructure which is designed to deliver video services from initial creation through to eventual consumption at the customer premises. The diagram outlines a high level view of a typical CDN:
Super Head End
In essence, this is where the bulk of video content is processed before onward distribution to the network edge and beyond (to the customer device). As such, there's a number of key activities that are found here, such as:
- Video Encoding - video can be found in a number of formats, from uncompressed HD-SDI (High Definition Serial Digital Interface) to the latest MPEG-4 AVC (Advanced Video Codec). The former will be running at a speed of 1.485Gbps, whereas the latter will be approximately 8Mbps. To make such a considerable bandwidth reduction, a phenomenal amount of compression must take place. Consequently, video headends contain encoders designed specifically for this processor intensive task, either encoding uncompressed video/audio or re-encoding video/audio to a lower rate.
- Multiplexing - service providers can receive content streams from a number of sources, which must be combined into a suitable MPEG-2 Transport Stream (the transport stream is a multiplex structure carrying one or more services, all multiplexed together). This is termed multiplexing and as such, the head end will often contain multiplexers designed specifically for this task. Multiplexers can typically cherry pick from incoming content, allowing the service provider to deliver services with multiple audio streams, alter ancillary data and also encrypt services as appropriate.
- Encryption - CAS (Conditional Access System) is the term given to the encryption process which ensures viewers can only access the content they have paid for. Consequently, the video head end is one place where CAS can be implemented. The only problem with carrying out encryption at the super head end is the fact that if you want to change the video in any way at a regional headend, the traffic needs to be decrypted and re-encrypted. ECM (Entitlement Control Messages) and EMM (Entitlement Management Messages) injected into the transport stream allow the subscriber to decrypt content and provide subscriber permissions respectively. Ad Insertion - Generally, video content isn't delivered to the service provider with adverts already included. The service provider can splice adverts into the program stream using a technique called Ad Insertion or DPI (Digital Program Insertion). This process is standardised, with SCTE353 standardising the way in which ad breaks are cued within a digital video stream, and SCTE304 standardising the way in which multiplexers interact with Ad Servers. Ad insertion can be carried out on a national basis at the super head end, or on a local basis in order to add local advertising. It is worth noting that in Europe, ad insertion is commonly achieved during content production, rather than at the head end.
- VoD Servers - Video on Demand services must support thousands of unicast sessions to individual subscribers. Therefore, VoD servers must have high storage capacity and have the ability to handle large amounts of connections simultaneously.
- Middleware - this governs how the subscriber experiences their service, from the look and feel of the EPG (Electronic Program Guide), to the timely fashion in which the subscriber should receive their service once they have asked for it. Middleware will also be present on the STB, and may also be used in the network to monitor quality.
IP Transport Network
The transport network is fundamental to IPTV distibution but it's important to remember that the network will not solely be transporting TV services. In today's converged networks, the the transport network carries everything. However, in the context of this article, two IPTV services need to be considered:
In this scenario, large volumes of the same content must be distributed at the same time, from the Super Head End (or Regional Head End) to potentially thousands of edge distribution nodes and ultimately hundreds of thousands of subscriber STBs (Set Top Boxes). If we set up a unicast stream between the SHE (Super Head End) and every STB, there will be a vast volume of duplicated packets passing between the routers of the IP distribution network. As such, IP Multicasting is used to deliver the content across the IP distribution network in order to provide an efficient delivery mechanism.
Protocols such as IGMP allow STBs to join multicast groups. In essence, as a user flicks from one channel to another, they are leaving one multicast group and joining another. This highlights a major difference between satellite or cable TV distribution and IPTV; in the former, subscibers receive all of the content all of the time, whereas in the latter, subscribers only receive the content on a program by program basis (as and when they join the multicast group for that program).
It's worth noting that PIM (Protocol Independent Multicast) is also found in the IP distribution network and is used as a mechanism to keep the multicast routes across the network optimized. This is on the basis that if nobody has expressed an interest in watching a particular program stream, there's no point in sending it across the distribution network. Hence, PIM will be used to prune the network to remove that particular branch, until the time that someone does want to receive the content.
With VoD (Video on Demand), subscribers can gain access to a menu of available programs which they can choose to watch as and when they see fit. This requires a different approach in the network, since the service provider will not be able to predict when and where a particular piece of video content needs to be delivered. As such, VoD services are based on unicast IP delivery, with video streams often being established using RTSP (Real Time Streaming Protocol). This protocol not only allows the STB to request a particular video stream, it also provides control of the video stream (Play, Pause, Picture Search etc). Therefore, with VoD, video is sent on a 1 to 1 basis between the Media or VoD Server and the STB. More often than not, VoD servers are found at the network edge in order to reduce the amount of traffic flowing across the core network. Thus, VoD servers at the Super Head End handle the less popular content, whereas VoD servers at the network edge handle the more popular content.
In Part 3 of the IPTV Instant Expert series, we will continue our route through the CDN and look at the activity that takes place at the Regional Headend and the Customer Premises.