The DVB is working hard on the next phase of OTT delivery, writes DVB Project Head of Technology Peter Siebert.
Over the past 25 years DVB’s standards have paved the way for the transition from analogue to digital broadcast, first in SD, then in HD and UHD quality. The DVB Project has now turned its attention towards making over-the-top (OTT) internet-based delivery of audiovisual content as user-friendly as our ‘classical’ delivery solutions, DVB-S/C/T and IPTV.
The new suite of specifications, to be collectively known as DVB-I (where the I stands for internet), will offer a similar user experience to linear broadcast TV and will provide an additional deployment option for broadcasters and operators.
In line with DVB’s usual approach, work on DVB-I began with the drafting of Commercial Requirements (CRs), now broadly finalised. In addition to providing some guidance on the technologies to be used for DVB-I, the CRs also consider classical broadcast solutions and explicitly allow for hybrid scenarios, where content may be consumed via a combination of broadband and broadcast access.
DVB-DASH, including its Low Latency version, will certainly be a core element of DVB-I, and also Adaptive Media Streaming over IP Multicast offers the potential for broadcast-like scalability in both managed and OTT networks. Technical work on these elements is already well advanced. The focus of new specification work will be on service discovery and the necessary signalling.
Technical work on service discovery is at an early stage. The CRs provide some pointers towards the likely components of the technical solution:
• Avoid binary descriptors; consider, for example, JSON and/or XML
• Use RESTful API designs
• Use well-deployed HTTP and HTTPS functionalities.
Some consideration is also given to the bootstrap mechanism, which is the starting point for locating DVB-I service offerings in different locations, territories and access networks, including cross-border service access. As one potential solution, the CRs include the possibility that a centralised registry and/or infrastructure would be set up. Of course, this is only one approach and a range of options will be supported.
When it comes to the other building blocks of DVB-I, the underlying technology is either already well advanced and implemented or in the final phases of standardisation. The first version of DVB-DASH was published in 2015.
It defines a profile of the MPEG-DASH specification dedicated to OTT delivery of broadcast content. As compared to the quite comprehensive MPEG specification, the profile of DVB-DASH facilitates implementation and conformance testing. Recently DVB published an updated version of its DASH profile, adding UHD and related advanced features, including XML-based subtitles.
“We’re expecting the DVB-I specification for service discovery to be available later in 2019. This will be a major milestone for DVB”
One annoying aspect of DASH delivery today is the extensive delay experienced when viewing live content, especially when key moments in sports events arrive later than the broadcast signal received by a neighbour. Here DVB is working on the Low Latency DASH solution. CRs have already been approved and are quite specific about the necessary performance.
For the encoder to screen latency, a maximum delay of 3.5 seconds is specified, which restricts the maximum segment size. Considering that each DASH segment must start with an I-frame, which has a significantly higher data rate compared to predictive and bi-directional frames, the data rate for a given quality will increase if the segments need to be smaller. To prevent an excessive increase in bandwidth, the CRs define an upper limit of 20% increase compared to ‘conventional’ DASH.
The technical group is now charged with identifying the sweet spot between these two boundary conditions, while still guaranteeing reasonable throughput and scalability of the content delivery network.
To achieve the required scalability for live broadcast content, DVB has provided a long list of CRs for Adaptive Media Streaming over IP Multicast. The technical specification will address managed networks in the first edition, before considering possible extension to OTT internet delivery.
In summary, the CRs focus on two points: the technical solution must provide broadcast-like scalability, without requiring existing clients (STBs, smart TVs, etc) to be replaced or modified. This is achieved in the network by adding a multicast server feeding multicast gateways, typically located in the home router, or at the edge of the Internet Service Provider network.
The multicast server in the network receives the media stream and encapsulates it into the delivery units of the multicast transport protocol. The multicast gateway joins the appropriate IP multicast group and converts the stream to a unicast signal to feed STBs or smart TVs in the home.
Since multicast streams are based on unidirectional UDP transmission, an additional mechanism must be provided to compensate for packet loss or transmission errors. For this purpose, the technical group foresees the use of both Forward Error Correction techniques and a unicast repair mechanism, which will correct or recover the original signal (if possible).
The technical experts in DVB are working on the specifications for Low Latency DASH and Adaptive Media Streaming over IP Multicast. Both documents are on track to be approved for publication in February 2019. We’re also expecting the DVB-I specification for service discovery to be available later in 2019. This will be a major milestone for DVB, bringing broadcast-like experience to the open internet.