Your browser is not supported. Please update it.

26 April 2006

Emerging Hybrid pay TV systems Part II ‘ Can hybrids be good enough and cheap enough?

Last week we discussed all of the extra costs that are needed to upgrade a telco network in order to make it ready for IPTV, with DSLAM upgrades, customers premises equipment, a QoS system, a new MPLS enabled aggregation network, and massively increased backhaul, not to mention a head end, VoD and encryption systems and the equipment that must be installed at a customer premises.

We made the case that pure IPTV triple play currently does not exist, and to deliver it to a large cross section of any society, will include costs that most commentators and operators have yet to fully factor in.

This week we want to answer the second half of the question. Can a hybrid be built cheaply enough to compete with all of this, and who, in their right mind, might want to build such a thing.

So the three key questions to answer are, Can hybrids be done well enough? Can they be done cheaply enough? And are there organizations in a position to opt for them.

Let’s talk a little about which hybrids are available to the largest US satellite TV company DirecTV. Satellite broadcast direct to home is the cheapest way of fulfilling a large footprint with multiple TV stations. There is some local broadcasting, but most of it is a single TV channel broadcast to an entire country.

The cost of the satellite dish and the set top combined, are roughly in line with the costs that IPTV has for set tops and other home customer premises equipment like wireless routers. Both DirecTV and EchoStar have recently added extra satellite capacity, which remains extremely cheap, in order to provide HD services. DirecTV last May lined up the satellite capacity to deliver 1,000 high-definition TV channels. That means it can deliver US wide HD plus local channels. Dish has done much the same.

They have each made profitable businesses and already have these components ready installed. What they now need to complete the triple play is VoD, telephony and broadband access.

DirecTV was the world’s first pay TV service to offer a DVR, through a deal with DVR inventor TiVo, so it has that also.

And a year ago, we highlighted a key invention from WiNetworks, an Israeli start up, which allows for the attachment of WiMAX antennas to the same cable used by the satellite dish receiver.

This cable can carry both pure IP signals up and down, and the DVB-S signal down, to the set top. A home network adaptor can also translate IP VoD signals for the set top, and also translate any signals downloaded by a PC. A similar return path network can be added to a satellite TV system using Wi-Fi or Wi-Fi Mesh, especially in a built up metropolitan area.

The backhaul problem is far smaller for this type of system compared to a full IPTV set up, because the huge multicast comes out of the sky. There is not one big 350 Mbps to 500 Mbps multicast delivery to find cable or, only each of the specific separate requirements of each household.

The costs are still there for a broadband remote access server, for a head end, for VoIP cpe, more or less the same costs that a pure IPTV triple play would have. Other IP based services could be delivered over the wireless broadband, or wi-fi, or whatever other broadband network is built out. Powerline could easily be used.

We’ve met with met Powerline specialists, one or two of which have told us that all of their opportunities are measured in millions of installs and that DirecTV’s Rupert Murdoch had been sitting down with them just a few months back.

Powerline tends to be delivered by Wi-Fi in the last 200 feet in the US because of the peculiarities of the electricity network there, using the mid-voltage power segment as the carrier around a city and Wi-Fi thereafter. This is because the US low voltage power segments only reach about 10 to 15 homes each and it would be inefficient to put a signal on them. In Europe Powerline is more viable using the low-power segment of the power grid, so that broadband (or a TV service) can come straight out of the power socket in the wall.

It’s not even necessary for satellite operators to provide their own broadband line. DirecTV has said that it will launch a VoD using ‘best effort’ internet by Christmas 2006.

Best effort internet is IP traffic that doesn’t have a guaranteed delivery rate using a quality of service protocol. Typically this will work in the way Akimbo works, a pay TV operator builds a DVR which can receive video packets in a file download ‘ a trickle feed ‘ which will eventually show up on its Electronic Program Guide when the video is ready to view.

But why would DirecTV go to all that trouble, and what kind of cost will it have to lay out?

In the US, DirecTV has a churn of 1.9% per month, that’s 22% per annum. DirecTV has an Average Revenue Per User (ARPU) of around $69 a month, more accurately $823 a year. This means that over 2.5 million customers leave DirecTV each year. If it could come up with a hybrid triple play, and if it could drop its churn to something more in line with natural wastage (the only people that cancel are those which move house, pass away or enter financial difficulties), let’s say 8.4% a year (0.7% per month) then its revenue would rise by $115 million for every 1% of churn it could eliminate each year. This is on top of how fast its revenue is currently rising at the moment. Right now it has to replace this 2.5 million customers each year, before it can begin to grow. And yet it does achieve growth.

In total if it could cut its churn figure to 0.7% per month, DirecTV would increase its revenues $1.5 billion a year faster than it is today.

It is perfectly feasible that it could also double its ARPU over a number of years from $69 to $138 if it could find a hybrid triple play. And that would quite literally double its revenues to something close to $24 billion

Since it costs DirecTV $626 to acquire a customer, of which $180 is unavoidable because of the provision of customer equipment, then it would ALSO save $446 for each customer that did not churn, in unspent costs. In total 1.9 million customers would not churn that currently do, and the savings would be $850 million a year.

Even if DirecTV NEVER improved its ARPU, some $2.35 billion a year would go to its bottom line in increased revenues and reduced costs in the first year that such figures became true.

Given that the calculated cost of building a US wide WiMAX network to provide DirecTV with a return path is around $3 billion, then it could pay for the network in 18 months if it GAVE away broadband service as long VoIP telephony was offered, and as long as it achieved the churn reduction due to the three services being extremely inconvenient to change, once they were signed for.

So using the example of satellite TV linked to broadband wireless, a hybrid triple play can certainly be done cheaply enough, because for some players it can be justified on churn reduction ALONE.

There’s another hybrid triple play that we know people are working on both in the US and further afield. This is a localized system using MVDDS (Multichannel Video Distribution and Data Service) to deliver IP traffic, including TV, again linked to any kind of minimal return path, such as broadband wireless. MVDDS is a one way, line of sight, spectrum.

This has just been through the second round of licenses in the US and it reuses 12.2 to 12.7 GHz satellite spectrum. A typical US city license costs around $9 million, and a total of under $200 million was raised by the FCC in two rounds of license that will cover virtually all of the US markets.

A single transmitter will deliver around 4.5 Gbps in an 8 mile radius. This is enough for 100s of Standard Definition TV channels, perhaps 70 to 100 high definition channels, plus it can be used as the delivery path for html files requested by a return path which can be wi-fi mesh, WiMAX or powerline or wired broadband or cable. It has enough bandwidth to Broadcast VoD, or rapidly fire the films across town at faster than real time, to be stored on a DVR.

If you add the cost of a local head end, plus the return path network, this presents a genuine economic challenge that small local telcos or local broadcasters could operate.

The weakness of this system is that it requires a line of sight tower, so perhaps a 300 foot high tower. It needs civil permissions for this, and not every building can get line of sight to a transmitter. This can be resolved by taking this to the top of apartment building and running it down the building using Powerline or Homeplug technology with the TV signal coming right out of your power socket. Of course a local Ethernet could also be used and so could wi-fi.

This also means that multiple homes could be clustered to share one antenna, which in turn drives down the CPE price. A system using this is going to need to receivers, one for MVDDS and one for the return path, let’s say broadband wireless in unlicensed spectrum. But if these are shared between several homes with either powerline or wi-fi delivering the last hop, this can be made economic.

We estimate that CPE could reach as low as $150 per home using shared powerline, and that includes all Powerline chips. At the moment a WiMAX CPE (currently $300 heading for $50 over the next two years) would need to be shared to make this economic, but only between two or three homes.

It’s complicated, but it will deliver a huge amount of TV to an entire city, for a cost of around $20 million, on top of a WiMAX or Wi-Fi network.

The inventors US operation, MDS America, plus Cablevision and EchoStar own virtually all the US licenses. We think that this would absolutely kill IPTV in a given area except for one thing, a single installation would have a lot of trouble getting hold of the content. But multiple implementations over a wide area, would bypass this problem.

Permissible uses for MVDDS under the FCC license includes any digital fixed non-broadcast service intended for reception by the general public, but not on a subscriber, basis which can include one-way video programming, one-way direct to home wireless data and two-way services may be provided by using other spectrum for the return or upstream path

Another very European hybrid at the moment is typified by the one that British Telecom is currently planning to roll out this year. BT has declared itself a Microsoft IPTV customer, but says that it doesn’t think that broadcast TV can be brought economically to the UK over its existing access network. BT worries that the huge investment that it is already making in turning its core network into pure IP ($20 billion over 5 years) is enough of a risk. Quite right too.

Instead, BT plans to ‘augment’ the existing Freeview service. Freeview is a service that is digital free to air, using a standard cheap broadcast aerial. This service uses DVB-T and has really taken off in the UK. It has climbed to 10 million homes in 3 years, while the pay TV services, BSkyB and NTL/Telewest using satellite and cable, have reached 8 million and 3 million respectively over a far longer time, out of a total of 24 million homes.

Currently Freeview is just 30 channels, but costs nothing to the consumer and is advertising paid for. The set top costs around $100 and is a retail proposition. By designing a set top that will allow a standard ADSL broadband line to plug into it and adding DRM logic, this allows a VoD service to be added to the system. In BT’s case the VoD service is made up of a huge movie store and all of last week’s broadcast TV. This type of augmented TV can hinder the take up of real IPTV dramatically.

The costs to BT are its existing broadband network, which barely needs upgrading. The design of a new, ultra cheap set top, which is a consumer proposition. A large server based VoD system, plus marketing and help desk. This is fantastically cheaper than a full IPTV service.

Again it will be done well enough, it will certainly be cheap and BT has already adopted the policy so we know that at least one company is interested in this approach. Across Europe we suspect this will become a ‘de facto’ offering in every market from one or other incumbent.

Another question remains over whether or not Broadband Wireless technology such as WiMAX, can offer a full triple play on its own. WiMAX has only been tried once with simple IGMP (Internet Group Management Protocol) but this has been shown to reliably deliver 20 Mbps of multicast video working in just a 10 MHz channel (in 3.5 GHz spectrum we understand). If this is used at either Sprint in the US, or at somewhere like Telenor (where we know it has over 20 MHz of such spectrum) or another of the global WiMAX licensees, it could easily deliver 20 channels of Standard Definition TV from the same base stations used for broadband data delivery. Qualcomm’s Flarion technology could do much the same.

Since the cellular protocols such as HSDPA and EV-DO are supposed to deliver high speed data to handsets, then as they reach the speeds they are supposed to in subsequent versions, up to 14 Mbps, then these technologies will open up the same kind of performance as WiMAX, if someone configures their network accordingly. So cellular could become a triple play route.

There are also likely to be many more 3G licenses sold in both Europe and the US, over the next few years, in some cases harnessing spectrum that was previously used by analog TV delivery.

These systems will have their own Multicast protocols in the form of the 3GPP approved MBMS and a similar system for CDMA. It has always been assumed that this would be used for mobile TV only, however there is nothing to stop a standard definition signal and potentially high definition, being sent across such a multicasting protocol using cellular, instead of targeting TV only for a mobile handset sized screen.

This includes systems like IP Wireless’s TDD implementation of MBMS and IP Wireless has told us that Sprint plans something like this in the future.

If you have to build basestations anyway then the increased cost of segregating a video IP stream and delivering it separately to each cell is measured purely in dedicating a small portion of the spectrum, upgrading the base station software and getting the multicast signal to more backhaul points, one per base station.

When building a broadcast network operators usually pick spectrum that will spread content far and wide, like DVB-T and DVB-H, usually under 2.0 GHz Spectrum. There are installations in 220 MHz, 600 to 700 MHz and 1.4 GHz and 1.6 GHz.

For interactive systems like cellular base stations or WiMAX, the picture on the right is more common. There are more cells, closer together and the cost is in the increased number of bandwidth points, and these each need separate backhaul. But then you have the basis of a triple play

How much bandwidth is needed for a hybrid Triple play versus a full on, High Definition pure IP triple play? The first diagram below is what Alcatel and its partners advise operators will be needed for a full IP triple play and is the target deliverable for AT&T. It anticipates around 8 to 10 Mbps for any kind of compressed High Definition stream, and three TV streams per household, with two SD and one HD for now. This will change over time as the H.264 and other codecs improve, as the MPEG 2 codec improved before it, until the same bandwidth will allow 2 HD streams and one SD stream. High speed internet (including voice) takes up an increasing amount of bandwidth as it is used more aggressively in home based applications. All diagrams are available in the PDF version of this article.

The second diagram shows that once the multicast segment is out of the way, the access network of a hybrid requires substantially less bandwidth per home.

Just 6 Mbps per home is required and the key to this is that when provisioning multicast video at the access level, every single access line is going to be busy at the same time, during prime time viewing hours. However when provisioning for IPTV services without multicasting, the only video that goes across it is VoD, and this is rarely in use by more than one in ten homes, concurrently. Operators could even offer HD films on VoD.

When people originally built out broadband networks they soon found that only 1 in 20 people were using the bandwidth at any point in time when using the internet.

So operators got used to under provisioning in the ratio of 20:1. With TV Multicasting they can no longer do this and need to provision for the full speed of the wire or ‘WireSpeed.’ That’s because a high percentage of the TV households are watching TV at any point in time.

When this falls back to VoD only, then, depending on how good your VoD options are, and what content operators have on their core broadcast TV system, only 1 in ten households tends to be watching at any point in time. This creates far less concerns for a network in terms of bandwidth, backhaul and QoS provisioning than if you offer multicasting. It’s an order of magnitude savings at the backhaul level and it reduces costs accordingly.

There are many potential hybrids out there and in order to establish what they might be, look at the two columns in the diagram below and mix and match any two, to find potential hybrids and apply the same criteria we have applied, in asking if they are cheap enough, good enough and if any major player is likely to enter that market. One thing we believe that is hugely underestimated is the interception of cellular services with triple play services, including Standard and High Definition video, given the roadmaps for high data rate mobile data services.

Next week we will conclude our series on the threat of Hybrid IPTV system on pure IPTV, by looking at the changing landscape of the entire advertising eco-system and other threats which may dent IPTV including downloading video files over the internet, and web TV services.