In the past six months, we’ve heard of streaming-related pains in DVB-H, and have been approached by some of the largest players in mobile broadcast to evaluate if Digital Fountain’s technology would prove beneficial in streaming video over DVB-H, with a specific focus on the most prominent pain points: infrastructure costs, QoS, legacy equipment support, and migration. The answers are all an overwhelming yes, and I want to preview the findings of a white paper that we’ll publish in late January leading into Mobile World Congress.
Digital Fountain’s research team created DVB-endorsed simulators to emulate a DVB-H system, and the results were astounding. Link margin gains using Digital Fountain technology (compared to the status quo) were 2dB – 8dB, depending on the accepted latency varying from DVB-H typical latencies up to several seconds too translate those dB gains into meaningful numbers, we were informed by DVB-H operators:
Using Digital Fountain’s technology, and assuming a realistic link margin gain of 5dB, would allow an operator to deploy a broadcast network with half the investment in transmission towers, or to achieve a substantially better QoS with the same number of towers. In an instance of say, Italy, where the DVB-H rollout has cost upwards of 300m EUR, Digital Fountain solutions could yield a start-up infrastructure savings of 150m EUR, and that’s gathered the attention of a number of operators developing DVB-H solutions.
So the immediate question we always receive: Why does DVB-H streaming not include Digital Fountain’s technology? This is also quite simple, because DVB-H specifications were completed before Digital Fountain’s technology was available. But with our flexible software based solution, it is never too late – upgrade today, save from tomorrow on.
Of course, the same principle applies to other networks as well—MBMS, DVB-SH, MediaFLO, DMB, et al. It’s an interesting development and Digital Fountain will be working with operators globally to share the data in the coming months.
Comments (1)
Interesting news, although link margin gains results should be always accompanied with the network latency increment.
I'm especially interested on the figures behind "varying from DVB-H typical latencies up to several seconds".
It is obvious that there is a significant potential gain in "transmission robustness" (see note 1) for DVB-H streaming services by performing multi-slicing forward error correction (i.e., coding several data bursts at the same time jointly), in opposition to the conventional intra-burst MPE-FEC coding scheme. The larger the number of bursts coded jointly, the higher the coding efficiency.
This technique can be easily implemented in DVB-H by performing application layer coding with Raptor codes.
However, the main drawback of this approach is an increase of the network latency proportional to the number of bursts coded jointly. This latency affects the user experience by delaying the initial reproduction of the services. In the case of mobile TV, this implies a larger zapping time between channels, which is currently seen as a crucial parameter for DVB-H usability.
Therefore, the trade-off that needs to be evaluated is the gain in robustness as a function of the latency introduced.
My investigations point out that relatively large latencies (from 30 up to 60 seconds) are required to achieve very significant gains, although interesting gains can be achieved for small latencies of around 10 s. My evaluations are based on vehicular urban DVB-H field measurements (see 2), and thus for lower velocities the latencies required would be higher.
I hope this comment initiates an open discussion about this zapping time issue within the DF community.
David
------------------------------------
David Gomez-Barquero, Ph.D. Student
Polytechnic University of Valencia
iTEAM Research Institute
Mobile Communications Group
------------------------------------
PS1: Note that the gain in terms of link margin is very difficult to assess in real life, and the approaches usually employed like for DVB-H file delivery services are rather simplistic.
PS2: I'm presenting the preliminary results at the European COST2100 action (www.cost2100.org) the first week of February prior to the Mobile World Congress. My paper is entitled "File Delivery for Streaming Content in IP Datacast over DVB-H Systems – an Overview". Interested people can drop me an email to get a copy of the paper.
Posted by David Gomez-Barquero | February 2, 2008 12:17 PM
Posted on February 2, 2008 12:17