We always go on about how tough it is that half of Europe can’t get its hands on decent mobile TV services, mostly due to a lack of spare spectrum for DVB-H or other broadcasting technologies, but there turns out to be more than one way to skin a cat.
A year ago a small US company called Broadcast International introduced a completely new concept in video over a constrained network like a DSL line or better still a switched cellular connection. The idea is fairly simple to describe, but tough to build, it takes 5 or 6 different approaches to digitizing video ‘ using different encoding algorithms – and digitizes and compresses video using all of them in parallel, and picks the outcome with the lowest bit count to use for each frame produced. Previous approaches used one standard encoding approach like MPEG2 or MPEG 4/H.264.
The improvement is radical because they have built specialized codecs to cope with bright light or dark situations and other specialist video conditions, and BI has managed to get a High definition TV stream being delivered in about 3 Mbps, which it says will shortly move to 1.5 Mbps, and a mobile TV stream over a cellular link, which is said to be every bit as good as a DVB-H signal, but using just 50 kbps. Existing H.264 HD efforts are closer to 8 Mbps and mobile TV between 250 and 500 kbps. The new technique could be equally applied to broadcast or streamed technologies, but with many of the major US and European cellular operators working on video constrained streaming set ups, it could be harnessed for almost nothing, compared to building a DVB-H network for a fortune.
Doing that amount of processing is quite a tricky thing to do, so BI had to go to IBM and use a blade made up of two IBM Cell processors, the same chip that drives the PlayStation 3, in order to find anything powerful enough, and yet cheap enough, to output all those video streams in real time. The Cell Broadband Engine is a chip which has 8 MIMD processors, which can each process multiple instructions at the same time, all reporting back to one master that organizes all the work. It took several $ billion to bring this chip to market.
Now, after an extra $15 million of funding, and spending more than 20 man years of hand tuning the system onto this complex parallel architecture, BI has the encoders ready and will show the finished product at IBC next month.
IBC is the International Broadcasting Conference in Amsterdam, which drags every technology business which addresses broadcasting and TV technology out to show their wares to more than 50,000 visitors from all over Europe.
Now the company also tells us that two wireless operators in Europe are already looking hard at using the technology to upgrade their streaming cellular TV efforts. Streaming TV on a handset uses up too much bandwidth to allow a decent picture, because if there are too many TV viewers all taking a piece of a base station’s bandwidth, that would prevent cellular calls getting through. In order to get a decent picture the handset needs to support QVGA, at 320 x 240 pixels, at 24 frames per second, otherwise it just looks blocky and pixellated, and people cannot watch it for more than 2 minutes at a time, and with the BI codecs 50 kbps is all that is needed to deliver such a video density.
The company has been trying to address the IPTV and cable market, with some success, but few operators want to go to the trouble and expense of replacing all their set tops, which are geared up for MPEG2 and H.264, and instead have asked BI to build a codec that will improve performance on the equipment that is already installed, and while this is working, it isn’t bringing quite so much improvement over existing systems.
BI CEO Rod Tiede told us last week, ‘Targeting MPEG2 and H.264 only, takes away some of our performance, we are managing to get full HD down to 15 Mbps on MPEG2 compared to 19.8 Mbps, but still get down to 3 Mbps on H.264.’
The company has HD at 1.5 Mbps working in its labs, but this position means that it has to build its system for existing set tops, then some time down the line the operator can switch out set tops, use the full BI CodecSys benefit and dramatically reduce bandwidth requirements at some point in the future.
Cellular broadcast mobile technologies like DVB-H take the strain off the cellular link, and provide a separate network which broadcasts at this quality. The problem with that it costs around $200 million or more to build for each European country, plus the price of any spectrum.
What the BI technology represents is a way for the Vodafones and Oranges of this world, which have strong mobile TV offerings, to drop the data requirement for decent pictures from 250 kbps or more, down to 50 kbps, which means they could support 5 times as many customers with their existing networks and on a handset that could all be done in downloadable software. Since Orange has gone on record as saying that it will have to stop signing up new mobile TV customers in France by the end of 2008 unless it uses another TV technology, this could become the godsend it is looking for, and the last nail in the coffin for the very expensive DVB-H.
BI said that it started out focussing entirely on IPTV applications and now has fresh interest in wireless networks from two companies in Europe, which it will visit while it is at the IBC show. The company will have a ‘proof of concept’ demonstration for wireless at IBC, and we could be seeing live services in mid 2009 if that goes well.