Your browser is not supported. Please update it.

6 February 2014

How rapidly will LTE Broadcast deploy across the world

If you ask a combination of Ericsson, Qualcomm, a few specialist software suppliers and listen to the handful of operators that are supposed to be in LTE Broadcast trials, you could get the impression that LTE Broadcast will arrive in style during 2014.

And that’s more or less what we tried to write up during 2013 as we researched a paper on evolved MBMS or LTE broadcast, whichever you prefer to call it. But digging beneath the news snippets, what you actually find is that this is the excitement of yet another new generation of writers who have no idea that we have been here once or twice before. And the business model remains largely unchanged and unworkable, for most MNOs.

The four names you hear most often associated with LTE Broadcast are Verizon in the US, Telstra in Australia, KT in Korea and KPN of the Netherlands, which has just begun its own trial. In each case there is a reason for past and future involvement in video, and LTE broadcast is not a technology that they can afford bypass without taking a long hard look. This was the list we found we could draw up 8 months ago, and with the not insignificant addition of AT&T, which we found in September was ‘known’ to be ‘investigating’ LTE Broadcast, nothing much has changed and no new names have been added.

AT&T decided it had to cover the Verizon move and said that it will use the unpaired spectrum that it bought from Qualcomm in 2011 in 700 MHz, to broadcast signals using its LTE base stations, spectrum the company spent almost $2 billion acquiring in 2011. So both Verizon and AT&T, supposedly spectrum starved companies, are both using ‘spare’ spectrum ‘ we wonder how many other MNOs around the world have that luxury.

The facts of broadcasting live TV channels to a mobile device have not changed during the past 10 years. You can do it in a number of ways, but to make it widely available to the broader cellular community you have to do it in a way that makes use of their existing infrastructure. Other efforts did not do that, and it’s worth a brief reminder of just what those efforts involved.

Initially we came across TV on a phone in Korea, and it simply had a new chip and antenna in it which could tune the existing UHF broadcast analog TV. The picture was awful even by the standards of 2002/3 when it happened, and work had already begun on things like T-DMB and DVB-H, which would offer a digital version of broadcast TV especially for a battery powered, QVGA sized screen that was then common on mobile handsets. The main reason was not that the picture was poor, as analog TV on a phone has continued as a gimmick ever since, but it was the poor battery life this gave to a phone, especially a 3G then 4G phone which already had trouble keeping up with its core connectivity and the modern, increasingly large, display in a smart phone.

When the first iPhone came out it doubled the average size of screens, and now the Phablet and tablet form factors have taken this up yet again. Around 50% of power in a phone is used backlighting the screen, and having the radio tuner on all the time also make another hole in the battery budget.

So systems like DVB-H, MediaFLO and S-DMB and T-DMB in Korea, all delivered a channel as part of a spread multiplex, where each channel arrived at the radio faster than real time, one after the other, and the device radio switches on and off and only takes the channel it is ‘tuned’ to. This saved about 90% of the battery used on the radio, but saved none of the battery used to power the screen. All these systems relied on coded OFDM, using time and frequency interleaving, all of which are highly effective in getting around fading or interference or multipath delivery problems, which is one of the reasons that OFDM is now used in most broadcast and two way systems from DVB-T to LTE.

But such systems only came to fruition successfully in Japan and South Korea and were partially successful in other parts of the world, such as China, parts of Africa and in Vietnam and the Philippines and places like Poland and Italy. Most mobile broadcast systems were tried and then switched off, some after they were launched, some never even got that far. It was a case of pay to build a network, then pay to get the handset makers to put a special chip in them, and then sell the service at a price deemed too high to make it worth it to the poor consumer.

But ISDB-T in Japan is used by in excess of 80 million people on an almost daily basis, which is why these markets were originally less impressed with iPhones. Korea still has a satellite system which is used by a few million people and a terrestrial broadcast system that is used by perhaps 25 million or more.

But each has a story about how it came into being and in each case either the local dominant broadcaster bullied the cellular community, and convinced them to source new types of phones, with new chips in them with a (slightly) higher bill of materials. Where a new network had to be built, for instance at MediaFLO in the US or at 3 Italia using DVB-H, the service did not make economic sense and eventually failed.

Where the existing broadcast network could be used (this was true of ISDB-T and ATSC M/H in the US) with cheap upgrades, it was at least possible to get the service off the ground, as long as a cellular operator ordered phones with the extra demodulator chips in them. Japan has always been an exception to most cellular rules and the 1 Seg ISDB-T phones were ordered and slowly adding that demod chip was much the same as putting a camera in each phone, the operator made no extra revenue, and met the bill of adding the component, and yet there was no discernible service revenue ‘ the service was free.

And when everyone around the world realized this, they found it made no sense to copy it. In South Korea S-DMB was a weak service with paid content, mostly imported, and would have died but it did not want to give the market to the rival consortium that built T-DMB. But T-DMB took off in Korea primarily because it had the first real time TPEG traffic avoidance GPS system built in, with real time updates to your personal navigator, meaning you could cut commute time in the Korean capital Seoul, by about an hour a day. The fact that it allowed broadcast TV was only a sidebar and first devices were all in-car devices.

China only took off because it wanted to use its own technology and it came up with a weird science project that involved an S-Band satellite signal being harmonized with a terrestrial signal, for in-building penetration ‘ this time the expense of building not one extra network, but two. In the end the satellite part of the equation was never used, after the chips were built to make it work.

These efforts made money for both US company MaxLinear, which dominated ISDB-T chip designs in Japan and for Siano Mobile Silicon, an Israeli pioneer in mobile broadcast. Very few other companies have ever made money out of broadcast mobile and in truth operators like AT&T and Verizon, did not try very hard.

The US MediaFLO service hit a wall, when it refused to pay real money for content, and instead took designated ‘mobile’ channels, that were minor subsets of real channels, with little action on them. Live sport was summarized in highlight programs days later, and channels like ESPN mobile were designed not to take viewers from the pay TV network ‘ deliberately hobbled, and yet costing more.

So we have gone full circle and we have to ask, ‘What is different this time?’ The technology is relatively cheap, although we would not like to put a full price on it. eMBMS comes from existing base stations, and yet at sports arenas around the US, Verizon plans to use extra spectrum that it has not yet built out, so you have to roll the cost of that spectrum into the financial equation.

The key issue of extra components for the devices has gone away. Ericsson has always said that there were so many reasons to have a broadcast capability in cellular ‘ such as broadcasting the most popular YouTube clips to your phone, or upgrading software over the air in a broadcast ‘ that it would come eventually in the standard chips. Qualcomm has not built it into the baseband, but had added it to the logic of the SnapDragon application processor, and smaller French player Sequans has followed suit. It only needs MediaTek to do the same and all future smartphones will be able to ‘view’ eMBMS broadcasts natively and this can be offered up for Apps to trigger.

Other app processor players like Nvidia and Intel will all fall into line if these larger market leaders offer this on their chips.

Still the base station software version needs to go through an upgrade, but this upgrade has many improvements in it, giving all sorts of reasons to move to it over time. In other words ‘ eventually ‘ all smartphones will be able to do this with an App and all LTE networks will eventually have the capability to switch on eMBMS ‘ and so we can see broadcast services emerging over a 5 year period, and the rollout is predictable, but not sudden.

There are other costs, but they are not huge. MNOs need to install a Broadcast Multicast Service Center (BM-SC) as well as an MBMS gateway. These are software components that can just be housed on general purpose servers and are a shared resource between a number of base stations in a region. The BM-SC can accept either native IPTV multicast or adaptive bit rate streams as input, using protocols such as MPEG Dash, and these can come straight from a CDN.
The BM-SC has the ability to fix any packets that are lost, and it then puts an appropriate forward error correction code on top of the wireless last mile delivery.
There is one server to many base stations. It’s not a huge cost, compared to
the potential service revenue.

Trials have shown that if you use 10% of a 20MHz spectrum slice, a system can deliver 5Mbps. So you have to decide how dense you want the video. If you offer standard definition, that might be 5 channels, or one channel in HD plus perhaps two more in SD. Or you can use a larger percentage of the 20 MHz for more channels or use a smaller screen form factor than SD and send far more. It’s all flexible. You can also concentrate service in one base station, a cluster or all of them. Some trials in the past have shown eMBMS working in a dedicated 10 MHz, offering 15 TV channels, but these were more like QVGA streams.

Companies like Verizon, which actually has live sports rights, such as the NFL on mobiles, will find a good reason to be early to market. But those NFL rights are not everyone’s cup of tea ‘ they cost $billions and other sports will cost more, once Verizon gets some play out of the money it has spent. It can either sell a paid subscription, but more likely it will give the service away for free for the key ‘high spending’ customers in both its TV FiOS service and its LTE cellular service, and get its money back through bundle upgrades, reduced churn and through higher customer numbers.

Verizon is actually using some non-deployed spectrum to deliver this system at arenas, in other words this spectrum is not on a base station yet, it’s spare, so it can use all 10 MHz of it or more. The configuration includes extra support antennas in the arena, perhaps Distributed Antenna System and Verizon is known to be delivering 6 different camera views of a single sports game, which you would just be able to flick through while at the game. Then it will deliver it out of market on mobile screens for a price, and then inside the same market if the stadium is full. But that’s just one use case, Verizon says it has 100 different use cases.

Faultline is currently running a series of 3 articles that we will release as a white paper later this month on sports rights, and right now the mobile element is seen as a bit of an extra helping of the cake. Eventually mobile rights will become the largest part of sports rights, and it will become a weapon of choice in bringing new customers to a network. But this migration is expected to happen slowly, certainly over more than 5 years.

That will see sports and other live rights become important for MNOs to get involved with – it will transition from being the property of the pay TV market leader into that of the MNO market leader. Verizon has a number of key plays in video including its Redbox Instant and the OnCue investment it has acquired from Intel, and when added to these sports rights it becomes a powerful weapon in attracting customers perhaps more rapidly than AT&T ‘ hence the desire by AT&T to keep up with Verizon, at least on the technology. Verizon plans to use video as a key differentiator in much the same way that AT&T used the iPhone, in the years that it had it on an exclusive basis.

But that begs the question of just how will a second string MNO like Sprint or T-Mobile competes using video. Quite simply they can’t do it on sports right investment power and so that leaves them less likely as candidates for LTE Broadcast, or more to the point, LTE Broadcast will enter their networks at some point, but not as the main differentiator for service. If they need to achieve a differentiator in video rights, they should look no further than partnering the cablecos, who have content rights, in some cases out of home mobile content rights, but no network to deliver it on.

But although Verizon has bought the TV rights of Intel’s OnCue, it still does not have a full 40 channel TV bouquet and is missing many key components for such an offering. For other offerings most existing TV channels will continue to be delivered from a web site, or one program at a time on YouTube, over the top, with no direct financial relationship between the video owner, the MNO and the customer.

And we have all seen that the easiest, quickest and cheapest method for delivering video in this way is using WiFi Offload. Very few video assets which are consumed over mobile need to be seen at the time they are broadcast, as a linear stream, which means most of this OTT traffic is on demand, and so there is not any prospect of broadcasting it.

There is a simple use case, that of sending the most viewed YouTube clips to all phones using a broadcast, which would cut down the OTT video dramatically, and especially insure against events like the Gangnam Style sensation, which slowed up mobile networks everywhere.

So most operators will continue to leave video outside the walled garden and simply tune their networks so that they can cope with the masses of OTT video that they are expected to deliver from the web to apps. Initially a handful like Verizon, KT, KPN and Telstra will offer new video services. If these are effective at either fighting churn or gaining customers, then others will dive in.

There is a natural delay in LTE broadcast in that the handset chips that are only shipping in volume now, need to be widely deployed, and that will take two years in the US, China and Europe, and longer elsewhere, so that it makes sense that one pioneer per market offers LTE broadcast services over the next two years, and only in markets that have widespread LTE availability. Then the most competitive second MNO will follow suit, with a walled garden offer, and the rest in any given group of operators in a territory will find they are forced to deploy the technology, alongside video offload, in partnership with broadcasters or pay TV operators, in order to fight back.

That makes LTE a slow burn, taking 5 years to build momentum globally. However with the focus on Verizon in the US market, other dominant suppliers will wait to see if it has any success, and then try to copy it rapidly and lead the chase to LTE Broadcast within their own markets.