‘Use Your 3G Spectrum. Stay in Control. Keep More Revenue,’ boasts the headline on the IP Wireless web page that tells its potential customers about TDtv. And that just about sums up why Sprint is taking such a good look at the technology and may eventually prefer it over WiMAX for use on its 2.5 GHz US spectrum.
Using an unpaired TDD band, like those sold in many European auctions as part of 3G licenses, would mean that Cellcos could use their existing spectrum for Mobile TV, instead of DVB-H, MediaFLO, T-DMB or a host of other datacasting mobile TV options, which require a separate network to be built.
Staying in control is a vital aspect for Cellcos, because, not wanting to invest in a second network, most Cellcos have partnered with broadcasters and equipment makers in the 40 or so mobile TV trials around the world, and as a consequence, find that their mobile TV will also be offered by competitors. Of course this scenario also means that the revenue from such a service is also shared.
DVB-H might as well put up a sign saying, ‘Use our spectrum, let us control the TV offering and share your customers with us.’
But DVB-H is still the platform of choice in mobile TV services, unless IP Wireless, a 140 man company which has swallowed up $230 million of Venture capital funding, can convince Cellcos to adopts its approach.
The approach is more or less as we described it last week, the use of a Time Division Duplexing unpaired signal, running the 3GPP multicast MBMS standard.
MBMS is a multicasting protocol where existing cellular base stations are used to send multicast TV signals, using the same radio to receive the TV signal as receives voice and other data services in the handset.
Talking to Jon Hambidge, VP of marketing at IP Wireless this week, he confirmed that this does require a new chip built for the handset and said that a standalone chip would be available at the end of the third quarter.
‘We need to add the ability to deal with a multipart bearer. You only need one radio because the TDD signal is in neighboring spectrum, but the phone can be doing two things at once, maintaining a phone conversation and watching TV, and it needs to be able to handle this. I mean what happens when a call comes in when you are watching TV on the phone?’
By March 2007 Hambidge hopes that the logic will be built into an existing handset processor from one of the existing suppliers. He wouldn’t say which company, but it would improve IP Wireless’ prospects if it was someone with a lot of market influence, like Texas Instruments.
IP Wireless, as the name suggests, deals entirely with IP and already has 35 trial customers with 15 of them moving into subscription services this year. They range from T-Mobile in the Czech republic, to trials with Orange in France, another in Slovakia, and services in New Zealand and 7 networks in Africa. The New Zealand service has been running for 3 years and offers mobile VoIP.
One problem with VoIP on a mobile is that packets get sent to a specific base station which has the radio connection to the phone, and then the phone jumps to a new Base station. All the packets that have been sent, have to catch up with new voice packets that go be a quicker route.
‘We’ve created our own mobile VoIP architecture, and treat the first site in any call as the home site. We create a virtual tunnel to get the packets swiftly to the new base station once a phone has moved,’ explained Hambidge.
The tunnel is a bit like an MPLS tunnel. The layer three routing information is stripped out of the headers, and a set of hops to the new base station are planned. A set up packets goes on ahead to these destinations and reserves bandwidth for the call and says to route all the voice packets behind it in exactly the same way. Without the packet headers needing to be read, the traffic runs much, much faster and the packets that were routed to the previous base station catch up.
But of course something very similar has to happen when a TV station is being viewed and the phone hops to a new base station. The channel being viewed may not have had anyone watching that channel and so it will have to initiate multicasting on that channel as the phone hops across.
‘We’ve also put in our own Quality of Service protocols to run video across the network. We define 8 types of traffic and 4 classes of user, so there are 32 types messages.’
So will IP Wireless get the deal at Sprint after the TDtv trial was announced last week? ‘If Sprint wants to move early on video, it’s only option is us, but that’s their call.’
What Hambidge is referring to is the fact that using 2.5 GHz any other way would mean either sticking with Qualcomm or going with WiMAX. The Qualcomm route uses a very different approach to video, with its own MediaFLO network, which would mean sharing the service with arch-rival Verizon. To go with WiMAX it would have to have a multicasting protocol written especially.
We know that Samsung has talked about (but not shown) adding video multicasting to its WiBro WiMAX-alike system, and Sprint is in trials with Samsung on this. Motorola is the other technology trial it has.
Given that Sprint has 100 MHz of spectrum in some markets and 60 MHz in most, it might easily use 5 Mhz of that, or indeed 10 MHz or 20 MHz, purely on video. The TDD slice it can use can be any of those, according to Hambidge, if it chooses IP Wireless. That would enable something more like 30 to 60 TV QVGA 30 frames a second TV channels
Similarly, Sprint could choose WiMAX technology to provide basestations, and yet still offer some form of multicasting, created in software at the base station. WiMAX operates with just over 1,000 separate multiplexed carriers, and it isn’t difficult to create separate traffic streams, or just allocate a specific amount of the packets to one Quality of Service controlled application ‘ in this case video.
This is what Samsung is promising to do for Sprint, but it may be beyond even Samsung to bring a mature, ready for market multicasting technology to the table, without a year of testing and another year to run out specialist base station software and client chips.
Already WiMAX is being beaten up by analysts for focusing too heavily on the mobility portion of what it might offer, pushing back the arrival of genuine WiMAX options further and further into the future. Adding video might push it back further yet again and risk it never happening.
Certainly Hambidge is aware that IP Wireless might be used as a bargaining counter, to show the WiMAX community that it is not the only game in town and to bring down equipment pricing. For that matter he says that Qualcomm is not out of the equation at Sprint, but similarly it may be using both WiMAX and IP Wireless to force Qualcomm to adapt its roadmap so that mobile video can be achieved in a number of ways, not just using MediaFLO.
One of the issues that Sprint is going to have to face, for any network design, is whether or not to plan for video straight away. ‘You get about minus 3dB when you add TDD to your cellular system. That means that you have to reduce your cell size to about 66% of what you would have if you did not plan to offer video.
‘But this can be done slowly. You can start out with TDD in just some cells, the urban busier ones, and coverage can fall away to EV-DO when customers are out of range. Video can still be delivered over the EV-DO, but only where there are far fewer cell phones being used, perhaps in rural locations.
‘You can gradually in-fill with more base stations as your video traffic begins to take off. There would be an upper limit to how much voice traffic can be supported while 15 channels of TV were also being sent over the alternative TDD channel, so you would bolster this by adding base stations and making the cells gradually smaller.’
In the end a Sprint network built in this way, when fully built out, would have 50% more base stations, which is not a huge cost for the addition of video to every customer, especially if it was added incrementally.
‘Once the network is built out, you can begin to look again at video. We can deliver traffic in a 1 Mbps stream over such a network, we could deliver 10 such streams. You might either deliver High Definition TV to the laptop or to your car, or use the higher density to extend the existing reach of standard definition cable TV to homes that are not passed by cable.
Sprint has cut a deal with a group of the most powerful cable TV operators in the US to deliver many of the existing favorite cable TV channels to mobile phones in the next two to three years.
Of course all of this would mean a complex architecture in the phone. A radio in each of 1.9 MHz and 2.5 MHz to have the fall back option of using EV-DO, but these are off the shelf, cheap chips and IP Wireless doesn’t think it is a problem.
In the same way that the Digital Video Recorder is changing the way US homes consume TV, a hard disk drive in a mobile phone, or much more flash memory, would lead to a similar change in the pattern of usage in mobile TV. Multicasting specific TV options through the night, when live TV delivery might be either reduced or suspended while customers sleep, could mean these programs are stored rather than viewed. A combination of the IP Wireless live multicasts, plus overnight stored programming, could provide both a rich entertainment stream and more opportunities to experiment. HD programming in the smaller format, Standard definition TV that can be played on a conventional TV screen, are two such possibilities and of course multiple creations of video carousels that deliver strands of specialist content ‘ sport, comedy, news ‘ to subscribers for first thing in the morning. ‘Phones will definitely become DVRs during the next cycle,’ agreed Hambidge.