Moving media: 5G and mobile video
The exponential growth in demand for video over mobile means 4G networks will eventually buckle. For 5G to provide the answer, does broadcast capability need to be built into the specification? Adrian Pennington reports.
With video-over-mobile traffic forecast by almost everyone to multiply exponentially in the next five years and represent 70-80% of all traffic by 2021, something, somewhere has got to give. Mobile spectrum is only finite, after all.
“At a certain point the existing 4G LTE technology will not be sustainable to cope with the massive growth in video data,” says Volker Held, head of innovation marketing at Nokia. “We need a new structure. This is the kernel of the 5G business case. Utilising it means we won’t need to talk about bandwidth constraints for the foreseeable future.”
The fifth generation mobile network promises much, but there is life yet in 4G, commonly known as 4G Long Term Evolution (LTE) despite considerable debate about which implementation of 4G matches the ITU standard. Candidates include 3GPP LTE which is often branded 4G-LTE, LTE Advanced (also known as 4G+) and WiMAX which is deployed by some carriers in the US.
“It is uncommon to see operators sell price plans that do not include 4G access if the plan has a data allowance,” says Tony Maroulis, research manager, Ampere Analysis. “There was a period around 2010 when AT&T, T-Mobile USA and a few other operators advertised DC-HSPA+ as 4G, even though it only achieved data speeds of 42Mbps. 4G LTE is a more efficient technology and it generally starts at 50Mbps, with the potential to reach higher speeds when more spectrum bandwidth is added.”
As an example, in France, Spain, Poland, Belgium, and Moldova, Orange’s 4G coverage has surpassed 80% of the population. In the US, Verizon covers 313 million people with 4G LTE, and its VZW 4G LTE network now spans 2.5 million square miles. Some 60% of data consumption over those services is video of some kind including support for YouTube.
“LTE Broadcast – multicast – was seen as a solution to the impending video demand, but it never materialised as the demand shifted from linear TV to video on demand, which requires an individual, unicast, connection,” says Maroulis.
4G networks support MBMS (Multimedia Broadcast Multicast Service), which enables broadcasts to multiple devices over cellular networks. MBMS allows a UDP style of broadcasting to mobile devices by reducing the need for a unicast connection for every device and reducing the bandwidth required for cellular data.
Can 4G cope?
According to handset manufacturer Huawei, video now accounts for over 50% of total traffic on many 4G networks. Each subscriber consumes an average of more than 3GB of traffic per month, rising 60% a year. It argues that network capacities must be expanded to accommodate more video service connections.
“With sufficient bandwidth and improved video compression a lot of the obstacles faced by current video could be solved,” says Maroulis.
Operators are optimising the 4G network using techniques such as ABR shaping and scheduled delivery. T-Mobile USA has launched a premium offer for its customers to receive unlimited video – provided that video is reduced to 480p and is from providers who agree to that limit – with apparent success.
“Adaptive streaming approaches, such as MPEG-DASH, also allow for a fine adaptation of the content delivery to the radio access conditions and the terminal capabilities,” says Gilles Teniou, senior standardisation manager, content and TV service at Orange Labs. He thinks the HEVC codec is a clear answer to bandwidth-constrained environments, enabling the delivery of video services in HD.
Since March 2015, a feature called MooD (MBMS operations on Demand) was added to 4G, permitting the dynamic switching between unicast and broadcast. This makes it possible for an operator to identify where and when users are watching the same content at the same time. eMBMS is then seen as a way to control the dimensioning of the network by efficiently offloading the network when needed.
However, video delivery remains the single biggest challenge facing mobile networks. EE forecast 80% of all traffic on the network will be video by 2020, and Matt Stagg, the operator’s head of mobile video and content says that maintaining the quality of mobile video in the face of demand is not just a challenge for operators. “This needs to be looked at end-to-end and new compression technologies can reduce the bandwidth required to deliver HD video by more than half,” he says.
In some places the bottleneck will be the backhaul from the mobile tower, notes Ampere’s Maroulis. In others the bottleneck will be the radio access network (RAN), when too many devices try and connect to the same tower. “Sometimes the bottleneck may even be the device’s modem – not all devices are capable of 150Mbps LTE-Advanced,” he says.
The LTE-B Alliance, founded by Verizon, EE, Telstra and South Korea’s KT in April, aims to push for all new smartphones to have the chipset and middleware capable of supporting the technology by the end of 2017.
Consumers are demanding higher quality, fewer delays and buffering screens, and video is not just limited to TV content and film, but a lot of social media is now video. Some 300 hours of video is uploaded to YouTube every minute, half of which is viewed on mobile devices, reports Huawei. In addition, 75% of Facebook video browsing is performed on smartphones.
The cost of producing HD screens has decreased drastically too. According to statistics compiled by Huawei’s mLAB, 77% of smartphones delivered in the first half of 2015 had a resolution of 720p or above. The standard 2K resolution on smartphones is 2560 x 1440, which is three times higher than 720p and Huawei expects that over 10% of new smartphones will be equipped with 2K-definition screens this year.
A premium 2K video experience will be one of the objectives of mobile networks and will become the mainstream requirement in 2018, according to Huawei. It even observes that smartphones equipped with 4K and 8K screens will soon be available.
“HD video is imposing great challenges on the LTE networks, which cannot yet fulfil all the requirements of 1080p and 2K videos,” it warns. “The capacity of a single LTE cell must be increased to accommodate more video connections, while data rates at the cell edge must be increased to meet the minimum rate requirements of video watching.”
Its version of LTE is branded 4.5G and, according to Huawei, can deliver peak rates ten times that of a single 4G cell. In December 2015, TeliaSonera and Huawei deployed the first LTE-Advanced Pro (4.5G) network with an outdoor peak rate reaching 1Gbps, claimed as the world’s fastest mobile network.
“Even if LTE provides significantly higher rates than the previous access technologies, the increasing demand for high quality content and the increasing number of simultaneous accesses to unicast requests – e.g. SVOD platforms – have to be considered,” points out Orange’s Teniou.
360° video services for virtual reality also promise a new immersive experience for which the required bit-rates are still to be identified (3GPP SA4, the codec group, is conducting a study on VR in mobile environments).
In addition, says Teniou, a high number of simultaneous unicast requests impacts not only the available bandwidth on the access network, but also the load on the cache or edge servers. “Getting the right video content closer to the user may become challenging in such a case,” he says. “A new compression format would help, for sure, under the assumption that it is optimised for the upcoming video services such as Ultra HD and VR.”
TV and video services are high-capacity services that require fast, reliable data connections. As such, they define the essence of 5G networks, which are focused on providing high-bandwidth, reliable and consistent network services with Forward Error Correction (FEC) capabilities.
“Video consumption is expected to be a continued driver in network traffic for years to come and video use cases are on the roadmap for 5G development,” says Adam Koeppe, VP of technology planning at Verizon.
The broad outlines for 5G have been agreed by organisations like EU 5G PPP (Public Private Partnership), initiated by the European Commission with manufacturers, telcos, service providers and researchers.
The specifications – some call them promises – include: regular mobile data speeds surpassing 1Gbps, peaks of 10Gbps (South Korea’s SK claim to have lab tested 50Gbps – 1,000 times faster than 4G), and a latency below 1 millisecond.
“While 4G was all about delivering data at high speeds, 5G will enable real-time, live video delivery,” says Arik Gaisler, senior director, product management, infrastructure, Kaltura. “This opens up lots of possibilities for real-time communication. We are starting to see live streaming gaining huge traction, with Facebook Live, YouTube Live, Snapchat, and real-time gaming platforms such as Twitch all investing significant resources in live video delivery. 360° video and VR – both for VOD and live – will also benefit from 5G, because VR relies on real-time data tracking and communication between the consumer and the service.”
However, 5G is not seen as a direct replacement for existing TV platforms and its use case goes far beyond media. “For 5G to support broadcast it will have to have an efficient broadcast mode which goes beyond the current broadcast features in 4G,” says Peter Siebert, executive director of the DVB. “5G will have to support broadcast features such as subtitles and the operator will have to fulfil coverage and quality of service requirements. The necessary tools to provide this have to be included in the ‘still to be developed’ 5G specification.”
Siebert believes that 5G sits at the peak of the Gartner’s Hype Cycle. “Even before the technology has been defined, various operators have already promised field trials and equipment manufacturers have given us the impression that 5G technology is just around the corner,” he says. “A network would have to be very dense with ample base stations, which need to be connected with high speed links to the network. An important cost factor will be the higher frequencies to be used that will result in a more expensive network and consumer equipment. All in all, building up a 5G network according to the current promises will be extremely expensive for the network operator.” The most efficient model, he suggests, is a ‘high tower – high power’ approach on which current broadcast networks are built: “It would make a lot of sense to integrate this technology in the upcoming 5G specification,” he says.
The hybrid proposition
Ericsson believes it critical for service providers to develop a hybrid approach using the technology that makes the most sense for the situation and type of viewing behaviour.
“We already know that in certain, rural parts of the world a mobile network can compete with fibre and meet the performance of even a fixed connection,” says Ericsson’s Gordon Castle, head of strategy area mediacom. “Fixed connections will never reach the majority of the world’s population so already we’re seeing that 4G LTE can be cost effective compared to installed fibre.”
Ericsson argues that for highly popular short/mid-tail content, broadcast technology continues to make the most sense as its very cost effective. For VOD and niche linear long-tail TV content, service providers can use broadband unicast to improve efficiency.
“Many devices are already connected to several access technologies, such as terrestrial/satellite antennas and fixed or mobile broadband,” Castle says. “By integrating content from these different inputs into one user interface, an improved user experience will be achieved and the content delivery costs will be reduced. LTE Broadcast will play an increasingly critical role in the future distribution of video to consumers, given that the majority of terrestrial TV is delivered via broadcasting, whereas mobile video is delivered through cellular networks using unicast through a separate video stream. Mobile networks can also provide broadcast delivery with the mobile network dynamically switching between unicast and broadcast, thereby optimising resource utilisation.”
As user behaviour evolves and technology advances, Ericsson expects to see a crossover point where broadband will prove to be a superior form of delivery compared to fixed-to-mobile and where mobile will have a greater impact in terms of reaching television.
It gives the example of in-home solutions that require high spectrum efficiency. “Globally, fixed broadband deployments lack the quality and capacity to provide a high level TV service and the majority of global households are unable to receive a fixed broadband connection,” says Castle. “New mobile technologies with higher spectral efficiency, directional antennas and portable home gateways optimised for video distribution, will reduce mobile broadband delivery costs. Evolved 4G and 5G technologies will play a crucial growth role in enabling mobile video to be a cost competitive alternative to fixed broadband and grow at a much faster rate than any other network traffic.”
In the US Verizon is trialling fixed wireless 5G use cases. Verizon’s Koeppe says: “While it is early in the development process, we expect there’ll be a need for additional hardware, cell sites, and other infrastructure. Part of the work at our innovation centres and with our partners will be creating much of this equipment – for example, developing even smaller cells with greater network efficiencies.
The availability of spectrum is key to the success and advancement of 5G technology. We are very pleased with the actions the FCC is taking to ensure that 5G moves forward and fully expect 5G to be an evolution of our 4G LTE network.”
A clutch of European telcos, including Deutsche Telekom, Nokia, Telefonica and Vodafone, say they will begin conducting large-scale tests by 2018, with a launch in at least one city in each EU country by 2020.
BT and Nokia are to collaborate on 5G use cases and trials, and to jointly develop 5G standards and equipment. The work will underpin the forthcoming rollout of LTE-Advanced Pro and 5G services by EE. Nokia is also already conducting trials of its latest 5G-ready radio equipment at BT Labs in Suffolk. Forthcoming proof-of-concepts will focus on technology enablers for 5G, including millimeter wave radio and convergence, as well as commercial applications of ultrafast mobile broadband, mission-critical services and the Internet of Things.
Ericsson says it has trials with 20 operators in the works and will collaborate with Cisco and Intel to develop the industry’s first 5G router. “This will be a critical addition, particularly when you consider the numbers of viewers streaming high quality video content on a daily basis,” says Castle. By developing the router, Ericsson wants to enable ultra-high speed wireless bandwidth and facilitate the growth of new internet-connected devices.
Some, like Kaltura’s Gaisler, believe 5G networks can ultimately replace legacy technologies such as WiMAX, DTT and legacy mobile technologies for broadcasting.
“And with fast, reliable, two-way communications comes the promise of further enhancement of the video streams: time-shifted TV services over 5G,” he says.
Ampere’s analysis is that while 5G could be used as a terrestrial substitute, it would likely require upgraded terminals – including TVs and set-top boxes.
“Additionally, if 5G was used to replace digital- terrestrial TV, then it would be competing with other data connections for bandwidth, unless it had a dedicated bandwidth assignment, which would make it not too different from the current set up,” notes Maroulis.
EE’s Stagg strikes a note of caution amid all the hype. “As a service [5G] needs to be looked at end-to-end to ensure an optimal experience and we cannot be complacent and file the requirement under ‘5G has lots of bandwidth to support 8K resolution’,” he says. “Another aspect we are addressing within the industry is ensuring that all video is not treated the same and labelled OTT content. Through our research and analytics we now know that live sport and popular events put completely different demands on the network and must be addressed with a one-to-many technology such as broadcast.”
BT and EE are also keen to look at convergence in 5G. To drive this forward, they are part of the consortium supporting the 5GPPP working group (5G-Xcast).
This media delivery solution will have built-in unicast/multicast/broadcast and caching capabilities, and it will enable media services to use any mix of the available mobile, fixed and broadcast networks, explains Stagg.
“The project will take a holistic approach in order to minimise the media delivery differences between the considered types of networks,” he says. “It will be end-to-end and cover everything from the physical layer and the radio interface to the transport and application layers including protocols and APIs. The project will contribute to the definition of 5G critical technologies and its standardisation in 3GPP with the development of 5G broadcast. This is a very important topic that has gathered very little, if any, attention so far, and is key to achieving the ubiquity, scalability and cost-efficiency required by the core KPIs for sustainable immersive large scale video.”