Multiscreen routes: problems and solutions to IP-based content delivery

Harmonic’s ProMedia Carbon: efficient transcoding is a key technology requirement.

From the straightforward business of delivering TV services to set-top boxes, pay TV operators now find themselves in the ultra-complex world of IP-based delivery to consumer electronics devices. Stuart Thomson looks at some of the problems and solutions in development. 

The ability to view on-demand and live programming on multiple screens is now seen as a standard feature for a growing number of pay TV platforms. For major pay TV players – particularly those in mature pay TV markets – it has become a must-have tool to secure subscriber loyalty and also simply to be perceived as a cutting-edge content provider.

This is a very different world to traditional pay TV delivered to operator-owned set-top boxes. Delivering content to tablets, smartphones and laptops means delivering via IP, typically over unmanaged networks, using unicast rather than broadcast technology. It has been a steep learning curve for pay TV operators that are accustomed to delivering video using tried and trusted technologies – and even for those telecom operators that launched IPTV services, using multicast technology, over the past decade and a half.

In addition to pay TV operators delivering a line-up of channels and on-demand content to tablets and smartphones, the combination of growing bandwidth, IP video and the opportunity to deliver to end-user purchased devices without the need to invest in expensive set-top boxes has fostered the emergence of new generations of ‘pure’ OTT pay TV operators such as Netflix that present a growing and disruptive source of competition to traditional players.

A raft of technologies have been developed to support IP-based OTT multiscreen delivery – and the key one is adaptive bit-rate streaming, which varies the resolution of video according to the bandwidth available at any given time.

Adaptive streaming has revolutionised the delivery of video over IP, but it has also created complexity in the form of a range of formats – HLS, Microsoft Smooth Streaming, Adobe Dynamic Streaming for Flash, Widevine adaptive streaming and now MPEG-DASH – and profiles that operators need to support.

Business models

In addition to the challenges created by this fragmented ecosystem, content and service providers are also highly preoccupied by the need to find a viable business model for multiscreen, and by what might be called the price of success, according to Boris Felts, vice-president of product marketing at Envivio.

“From a technology point of view there is still the debate about the best architecture and how to make it scale. One big fear about unicast is that the more successful you get the more traffic you are going to generate,” says Felts. “So the [next] phase is about economics and paying for the growth of services.”

Felts says that the ‘success’ of streaming services during events such as the Olympics mean that broadcasters have to pay a premium to the CDN providers that deliver their content around the globe. And they often need to do this without gaining any real additional revenue from the service they are providing.

“There is a need to find a model that compensates for these costs. There are two ways – paid subscriptions like BSkyB’s model or the example from the US market where you have targeted ads in OTT streams,” says Felts.

However, targeted advertising is very much still in its infancy. And for existing service providers that are deploying pay models for multiscreen offerings, there is a danger that things like advertising-based OTT services, provided separately and unbundled from the main pay TV offering, could damage their tried and trusted subscription services.

For these and other reasons, pay TV providers remain ambivalent about pushing OTT and multiscreen services too rigorously. Some service providers with their own fixed-line infrastructure view multiscreen as a way to sell additional bandwidth – France’s quad-play operators for example are using content as a way of showcasing the capabilities of 4G mobile services. Operators in the US meanwhile are more likely to view multiscreen services as a way to engage viewers with enhanced and targeted advertising.

RGB Networks’ vice-president, international, Simone Sassoli points to the growing importance – especially in the US – of ad insertion technologies for multiscreen TV.

“There are now more eyes watching and deployments with a large number of channels that are ad-supported. The opportunity to replace ads is there,” he says.

This means there is a need for technologies that can support the insertion of adverts into services including network-based DVR, encoded for adaptive bit-rate streaming, allowing operators to replace ad content stored inside the CDN that has become out of date.

To help with this the US CableLabs organisation has developed standards that can work in tandem with the existing SCTE 35 and 130 standards to deliver advertising insertions to TV services that are delivered over IP networks: the Event Scheduling and Notification Interface (ESNI), which allows programmers to notify operators of upcoming alternative content; and Event Signalling and Management (ESAM), which specifies the API for processing real-time signals.

There is still, however, a major hurdle to the  potential of advertising revenue being fully realised: the need to be able to track who is watching the video and when. There is a need for integration with audience measurement systems so that back office infrastructure can record when content has been watched.

Advanced functionality

Whatever the business model chosen by operators to make money from services, users are beginning to expect that multiscreen offerings delivered by pay TV operators should match the experience they have when they take a mainstream pay TV service. “In that context a much denser solution and total cost of ownership become more relevant,” says Sassoli.

While Envivio’s Felts says there is “not a big technical challenge in providing software stacks for the majority of smartphones”, the fact is that consumers increasingly expect TV to their phones and iPads to at least match the functionality of TV services delivered via set-tops. It is not always easy for service providers to meet this expectation. For example, the diverse range of device-based players in the market offer, says Felts, “poor support” for things such as captioning and subtitling – a particular problem in Europe, with its multiple language requirements.

The notion that mobile TV should match the quality of broadcast TV also means that viewers increasingly expect TV everywhere services to deliver high-resolution premium content, which raises security concerns and leads to the creation of new rules designed to protect revenues.

Ludovic Pertuisel, video networks product manager at video technology specialist Thomson Video Networks, concedes that the coexistence of multiple standards makes it very difficult to provide the same features on second screens that viewers are accustomed to on their main TVs. Take the question of captions. Standards to support captions do exist in the world of web video – notably W3C standards WebVTT for plain-text subtitles in HTML5 video, and TTML, which includes a set of Distribution Format Exchange Profiles (DFXP) and the Society of Motion Picture and Television Engineers’ extensions, including SMPTE-TT, encompassing additional features for captioning – but Pertuisel says that these together are not really adequate. He says the HLS format – by far the most commonly used adaptive bit-rate technology – does not support subtitling standards. It consequently remains difficult for service providers to deliver an experience on tablets and other devices comparable to the one that viewers are familiar with on TV. [icitspot id=”123871″ template=”box-story”]

Ian Trow, senior director of emerging technology and strategy at video technology provider Harmonic, points out that the delivery of subtitles in interlaced content – the standard still commonly used for broadcast TV services – took “years to master” even without the issues of different frame rates and screen sizes associated with multiscreen services. However, solving this issue for multiscreen will be important not just for accessibility and regulatory reasons but because for many services the ability to insert text and graphics is crucial to the business case. He cites the example of shopping channels, which will be unable to build multiscreen business models without the ability to delivery text as part of the service.

Adaptive bit-rate standards

There is no question that adaptive bit-rate streaming has been a massive boon to those delivering video over IP to any device. However, even leaving the issue of captions, subtitles and graphics aside, the absence of a standards-based approach and the range of bit-rate profiles and packaging formats that content providers have felt obliged to support – not to mention variations for multiple devices with different screen sizes – has created a degree of confusion and even chaos in the market.

“Lots of customers have had a bruising time. People expect content to be available in multiple formats, and broadcasters have not been able to make back the investment,” says Trow. “The first wave [of operators] to implement multiscreen were pretty gung ho – they were cutting their teeth and trying to achieve a technology first. Now they are trying to recover some kind of margin and hoping that standards bodies work like MPEG-DASH will create some sanity in the market, so there is some commonality with the infrastructure that exists for linear programming.”

MPEG-DASH – established as a standards-based approach to adaptive streaming – has held out the promise of reducing the number of profiles that content and service providers have to support by reducing the number of aspect ratios and DRMs required to reach the most commonly used devices. Currently, says Trow, operators need to deliver video in 32 different profiles to reach the 400 or so most common multiscreen devices. MPEG-DASH holds out the promise of reducing this to between six and 12 profiles.

However, a number of challenges persist with MPEG-DASH. “Standards are only useful if they gain traction,” says Trow. There are still relatively few MPEG-DASH-compliant devices and, even if more become available, there remains a danger that the proliferation of the standard could result in its dilution. “DASH could try to be all things to all men,” says Trow. “We as a company are fully behind it, but the take up for consumer premises devices has been slower than expected. It would solve problems but we are not commercially naïve about DASH.”

DRM systems

Technology experts broadly agree – though to differing extents – that MPEG-DASH could solve a range of problems. David Leporini, executive vice-president, marketing, products and security at content security and interactive TV specialist Viaccess-Orca, maintains that MPEG-DASH could confer significant advantages not only by reducing the number of profiles that operators need to support but by clearing up the complexity of combinations of adaptive bit-rate formats and various DRM systems. “MPEG-DASH could ease the DRM/delivery protocol combination complexity,” he says, pointing to the example set by the DVB standard in the past, which enabled conditional access systems to be plugged into traditional broadcast pay TV platforms with relative simplicity. “MPEG-DASH will do more than standardise the protocol – it will also standardise the way DRM is integrated with the delivery protocol,” he says.

Leporini is optimistic that MPEG-DASH will achieve fairly rapid take-up. “I think deployments will happen in six to 12 months,” he says. He believes that the combination of elements of OTT delivery, including MPEG-DASH with HEVC encoding, will become standard, driven by the need to support an ever-growing range of OTT services.

Dan Peters, director of product management at content security specialist Saffron Digital, on the other hand, stresses that the device universe is still fragmented and challenging from a content security perspective. In that context, operators are likely to focus on delivering multiscreen services to a relatively small number of popular devices rather than worry too much about being present on every smartphone. “About 20 devices have an 80% market share [between them], so for premium content you are talking about a maximum of 20 devices,” he says.

However, the situation is further complicated by the much shorter product lifecycle of tablets and smartphones compared with set-top boxes, according to Peters. Whereas pay TV operators built their business model around a relatively long hardware upgrade cycle, mobile devices and PCs tend to be upgraded more frequently, he says. This can raise the cost of content security implementations – even more so if the providers pay DRM licences on a per-device basis for devices that may not even be used to access the service they provide.

“When you are charging a per device cost that can become expensive for service providers. For on-demand and catch-up for premium content you need to use a DRM and you could be in a scenario where even when you have people who are not paying for subscription content, you are still paying DRM licences for everyone,” says Peters.

To help smooth the path to realistic models, he points to Saffron Digital’s own strategy of implementing DRM within apps on the Android platform rather than trying to ensure DRM is hardware-enabled in every device. Another initiative designed to simplify the rollout of premium services – in this case by making common hardware-level security available across devices – is the Trustonic joint venture created late last year by UK chipset giant ARM in partnership with two other firms, Gemalto, and Giesecke and Devrient.

There is no doubt that the multitude of devices with relatively short shelf-lives that operators want to address makes life difficult. One possible way forward is to  do as much as possible in the network or cloud and reduce reliance on the client to enable the service.

Envivio’s Felts believes that the adaptive bit-rate world is beginning to coalesce around the most popular standards – with HLS out in front by a good measure –  and he is sceptical about the potential of MPEG-DASH to iron out the problem of fragmentation. “Inside DASH you have a few flavours you can pick from. When you say you choose DASH it just gives you access to tools but doesn’t guarantee interoperability between devices,” he says.

Felts believes that one trend that can simplify implementations for operators is to pull back as much functionality as possible to the network rather than trying to implement everything in the player on the device – ad splicing and DVR being two examples of applications that could be done more efficiently from the cloud. “You can do most video things from the network,” he says. “It’s a trend we are trying to push on our side – it’s hard for any one provider to build a solution inside the player.”

Hunger for bandwidth 

Simon Frost, head of TV marketing at Ericsson meanwhile says that MPEG-DASH is “looking promising” but admits that the standard is unlikely to be “a panacea”. He thinks that client support for the standard in the form of generic players will boost uptake by service providers, as will a wider embrace of the technology by key players such as Netflix. One issue is that multiple devices using multiple standards can effectively compete to deliver the best-quality experience to their respective users. “Those algorithms can be greedy in trying to achieve maximum quality. If you have a lot of devices on WiFi the way they work is try to achieve the top bit-rates all the time,” he says. Ericsson is working on a concept tentatively entitled ‘weighted fair network queuing’ which could be applied in gateways or DSLAMs or, in the case of mobile networks, quite close to where traffic is handed over from the fixed to the radio network. The aim is to apply patented techniques to “curb the enthusiasm” of devices to achieve maximum quality at the first available opportunity and hog the available bandwidth at the expense of the experience on other tablets and smartphones in the home or mobile cell.

RGB Networks’ Video Multiprocessing Gateway enables multiscreen delivery.

The shortcomings of adaptive bit-rate technology have implications for bandwidth management by service providers. If players on devices seek to display content at the highest bit-rates possible, the service or content provider may have to pay the same CDN costs whether there are many or few people watching – when the network is quiet the players will eat up the available bandwidth by playing back video content at ever higher bit-rates.

“OTT is unicast and the more popular the content the bigger the bill,” says Thomson Video Networks’ Pertuisel. “But if you consume at the highest bit-rate possible, and if the service provider doesn’t have its own infrastructure, the CDN will charge according to bandwidth used.”

Pertuisel on the other hand believes that the arrival of MPEG-DASH will “simplify the landscape by reducing fragmentation and accelerating adoption on many devices”. He says Thomson is receiving a growing number of requests for DASH implementations.

There is, to be fair, a strong wind behind MPEG-DASH’s sails. Samsung supports the standard for its connected TV platform. MPEG-DASH has also been adopted as part of the HbbTV standard for hybrid broadband and broadcast services in Europe. But HLS remains the dominant format for now.

However, the lack of a clear certification path for MPEG-DASH and the absence of a reference client, among other things, means that the standard is likely to be of limited use in straightening out the complexity of the transcoding piece of the delivery chain for now, according to RGB Networks’ Sassoli.

According to Sassoli, one key trend is for the separation of transcoding of video to multiple bit-rates in MPEG-4, for example, from the ‘packaging’ element – the selection of a particular adaptive bit-rate format. This enables service providers to distribute mezzanine format copies of video files close to the edge of the network for onward packaging – on the fly – in HLS, Microsoft Smooth Streaming or DASH formats when they are required. “For long tail content it’s appropriate to store mezzanine copies and when a request is made you adapt it to the format,” he says. “It comes into its own with customers deploying network DVR and VOD services.” This ‘just-in-time’ packaging also enables DRM to be applied on the fly for different types of delivery networks.

Sassoli says techniques like this are likely to become more important as the capabilities of devices increases. With tablets and other devices now supporting full HD resolution video and set-top boxes increasingly supporting adaptive bit-rate streaming natively, service providers are likely to be under pressure to deliver even more bit-rate options in the future in order to deliver the optimal viewing experience to match both the available bandwidth and the capabilities of the device.

“Multiscreen is not niche any more,” says Sassoli. Video services are increasingly IP-based and some greenfield operators are using adaptive bit-rates for their main TV service, he says. This also means that operators are looking to deploy “converged headends” with the capability to encode both interlaced and multi-bit-rate progressive video streams.

Back-end infrastructure

The complexity created by adaptive bit-rate technology is one source of additional costs to service providers – even if that cost is more than compensated for by the improved quality of experience afforded to users. But the operational cost of deploying multiscreen services also remains high.

According to Ericsson’s Frost, service providers are struggling to define a common look across devices in a way that is economic to operate and efficient to manage. In their drive to achieve consistency while reaching a large number of devices, content providers have invested in ‘silo’-type back-end infrastructures that will be difficult and costly to maintain as services evolve and become more popular.

“Right now the operators are putting sticking plasters on a complex back end,” says Frost, who adds that the development of HTML5 and the simplification of client software is paving the way towards cloud-based implementations that could simplify things generally.

“HTML5 simplifies the whole presentation aspect. The industry is looking to lower the cost of the device itself. There is a shift to more browser-based implementations for the client and more processing or playout in the cloud…just using a browser to render the user experience,” he says. However, there is still some way to go. “What we see is there is still an expense in unifying those experiences at the back end as a result of siloed investments. When competing operators start to enhance services and functions there will be rapid consumer evolution and how to keep pace is going to be a challenge,” he says. Providing unified billing for multiple users at a single address is likely to test back-office systems, but it is one that key players are now beginning to address.

However, given the extent of legacy infrastructure out there, only piecemeal migration to a pure IP platform is likely. Operators will continue to deploy hybrid systems, using broadcast technology to delivery mainstream linear channels. Pay TV operators are also likely to remain wedded to the set-top box and the branded experience it supports for some time.

Another challenge to cost-effective delivery of content to multiple screens raised by Frost is that of access to adequate infrastructure. Pay TV operators and other players without fixed-line or mobile infrastructure need access to third-party bandwidth to delivery multiscreen services. “Getting access to networks mobile or fixed is going to be key,” he says.

Bandwidth is the single most important cost for operators. Harmonic’s Trow says that with video traffic expected to quadruple over the next three years the problem is only going to get worse. “A lot of OTT services have not been able to achieve broadcast quality,” he says.

In the case of live services, the availability of bandwidth is particularly crucial, especially for high-profile events such as the Olympics that require millions of concurrent streams to be delivered. One solution is to introduce multicast technology to as close to the edge of the network as possible, with transcoding to adaptive bit-rate formats performed at a point where the service becomes unicast, as enabled by technologies such as Envivio’s Halo.

Envivio’s Felts points out that at one extreme, service providers could multicast to the home and transfer to unicast in a gateway or cable modem. However this is costly in terms of maintaining the gateway devices and ensuring compatibility with new formats. It also fails to solve the problem of delivering video to people moving outside the house.

At the other end of the spectrum is a traditional CDN architecture, without multicast technology but with caching of popular content at various points in the network. Unicast in general is a much more expensive delivery technology than broadcast for popular content, but there is little sign that the kind of multicast technology used over the last decade to deliver IPTV services will be widely adopted in the OTT world.

Ericsson’s Frost nevertheless says that bringing additional intelligence to the network can pay off in terms of operating savings. Ericsson earlier this year launched its LTE Broadcast technology platform, which combined the eMBMS multicast standard, HEVC encoding (see sidebar, p.8) and MPEG-DASH adaptive bit-rate streaming. A key appeal of the technology, says Frost, is that the network can be configured dynamically for multicast or unicast depending on demand. The challenge with the old MBMS multicast standard is that it required dedicated capacity. “If it wasn’t used you lost money,” says Frost. He says Ericsson has seen a “huge amount of interest from operators with LTE networks”, including Verizon and Telstra, with whom the company has signed deals. “As soon as you can guarantee quality of delivery, that has value for content owners,” he says.

That quality of delivery is clearly the key in appealing to consumers and in enabling service providers to make money – whether from subscriptions or from advanced advertising. But the lack of standards to enable all the advanced functions that people are used to from traditional TV services, combined with the high cost of supporting a bewildering range of formats and device types, concerns about security and the cost of delivering unicast streams over CDNs, all mean that the vision of economically sustainable OTT video remains beyond the horizon. For now, the world of web-based multiscreen video delivery remains a wild and untamed terrain compared with the cleanly manicured walled garden of traditional pay TV.

There are clear signs, however, that technologists are busy working on solutions that will reduce the barriers over time.