Mirroring the enthusiasm of users globally, India’s live-streaming market is set to explode despite numerous technological bottlenecks
By Satyam Nagwekar
Technology has transformed consumer habits at a geometric rate. In the age of instant gratification, the desire to witness even the smallest of events as they happen has been fuelled by live-streaming technologies.
While YouTube and Facebook added live streaming to their feature set a while ago, the former recently shot up to the top of the App Store–a position it hasn’t held since December 2015–following reports of the video-sharing site opening up mobile live streaming to more users. YouTube has relaxed its earlier restrictions on mobile live streaming, which was initially only available to creators with 10,000 or more subscribers, down to 1,000 subscribers. Digital video streaming, according to Sony Pictures Networks India’s EVP and Head (Digital Business) Uday Sodhi, is expected to be the next wave of digital consumption.
Ericsson ConsumerLab 2016, the global TV and media report, stated that average viewing time on mobile devices has grown by more than 200 hours a year since 2012. Moreover, 40 percent of consumers globally are ‘very interested’ in a mobile data plan that includes unrestricted video streaming.
The Indian market has largely kept pace with global trends. Said Sodhi, “The video-viewing audience in India is expected to expand significantly. The number of smartphones in India crossed 300 million during 2016. Even though majority of wireless internet subscribers are 2G users, the adoption of 4G is gradually increasing.
Right now, 3G/4G constitutes over 50% of the overall wireless internet user base. As it rolls out, 4G is expected to contribute to around 58% of the mobile data traffic in India by 2020.” He added that the digital advertising industry stood at INR 6,825 crore and was expected to increase at a CAGR of 37% to reach INR 23,795 by 2020 due to the increase in mobile phone adoption along with increased internet usage.
Abhishek Sood, product director at Accedo, believes that with the demand for streaming quality rapidly transitioning from SD to HD quality and moving towards 4K, audio and video compression technologies have a big role to play. “How the current compression technologies, such as HEVC, VP9, MPEG-H, and AC-4, and the new ones evolve over the next few months will decide the uptake of the live-streaming services among consumers. Without continuous and significant improvements in compression technologies, it will be hard to tap the live-streaming potential, especially in growing APAC countries due to network challenges,” said Sood.
Sports and breaking news are key drivers for mass adoption. Live streaming over IP networks is already a reality. “IPTV and cable TV operators have been using the feature for a long time on main screens connected to set-top boxes. It has been extended to other screens thanks to adaptive bitrate streaming technology. A growing number of end-users now consume live feeds on tablets, smartphones, and connected TVs,” Nivedita Nouvel, VP (marketing), Broadpeak pointed out.
Another trend is access to cloud-PVR features. Recording live content and playing it on any device is also an innovative service enabled by IP technology.
“Low-latency encoding optimised for packet-based IP delivery will boost the acceptance of live video streaming. Real-time optimisation in this environment is a challenge, especially if the buffer model and frame rates are on par with DTH applications,” Ian Trow, senior director (emerging technology & strategy) at Harmonic, said. Another issue is making delivery scalable for live streaming.
Multicast allows streaming to mimic DTH broadcast but widespread deployment is far away, particularly for OTT service providers that do not control the delivery infrastructure. “CMAF (Common Media Application Format) chunks and HTTP/1.1 chunked transfer make live streaming more manageable,” he added.
According to Research and Markets, the OTT market has grown at a tremendous pace with revenue slated to reach $18 billion in APAC in 2021. Moreover, a recent study released by Ooyala reports that 51% of all video is viewed over mobile devices presently, a 203% increase from 2014.
“Operators need to invest in mobile-first services in order to offer high-quality, live-streaming services. Delivering HD video to mobile devices is, however, proving difficult with the bandwidth constraints of 2G networks. In emerging markets like India, this prevalence is even higher with majority of the population relying on 2G networks,” said Fabio Murra, SVP product & marketing, V-Nova. Next-generation video encoding solves this bandwidth challenge, delivering quality pictures at the bandwidths that matter, without requiring expensive investment in infrastructure.
According to Nouvel, there are two key features that live streaming to multiscreen lacks right now. “One issue is scalability to manage consumption peaks during high-popularity events. With unicast delivery, the only mode supported by new devices, bandwidth resource consumption grows linearly with the number of viewers. Another problem is latency during live streaming.”
These drawbacks can be resolved using multicast ABR technology, which encapsulates the ABR unicast stream into multicast in the headend to convey it in the backbone network at a lower cost. The stream is then converted back to unicast in the home network. The resources used in the backbone network are kept low since they do not vary with the number of viewers and it is completely transparent for the device.
There are a number of large OTT services offering HD-quality live streaming but the ecosystem needs to act in unison to expand reach. “The source content needs to be captured in HD and encoded with robust DRM, the audio and video compression technologies need to evolve to reduce the packet size, and the carrier networks need to be able to carry HD and 4K live video. So, while there have been improvements on all these fronts and the CDNs and DASH have improved streaming quality, the demand for HD live streaming and 4K content will continue to challenge the OTT industry,” said Sood. In order to find a robust solution and overcome these challenges in the long run, improvements in carrier networks, compression technologies, and scalable delivery by CDNs would be imperative to achieve the common goal.
The quality of experience, latency, and data costs for the user are the three primary challenges for operators looking to offer live HD video streaming.
“The widespread use of 2G networks and the global average internet speed of 6Mbps, combined with numerous data drops recorded by the Telecom Regulatory Authority of India, are creating headaches for companies providing live streaming. This is particularly problematic for operators looking to deliver monetisable HD video. The market, and Netflix recently, agrees that streaming needs to be delivered at 300 kbps to offer good quality of service (QoS) on mobile networks,” Murra pointed out.
By deploying new video-encoding technologies, operators can significantly lower data costs and buffering time for the end user, making movies and live sports a mass-market offering, he said.
The IP Factor
IP-based infrastructure has made significant inroads to rationalise production infrastructure but, for streaming, multi-profile delivery necessitates the flexibility of IP-based infrastructure. “CMAF opens up the prospect of low-latency OTT streaming to challenge what is available in a traditional DTH workflow. While low-delay modes have been available in previous streaming standards, CMAF aims to improve upon these, reducing latency by 40%. Other benefits of CMAF are the replacement of the MPEG Transport Stream, which has significant processing overheads, and the simplification of processing through the emergence of one media format,” Trow stated.
“I think IP infrastructure has been a big revolution in the video delivery space and has made all other things possible. The packet-based delivery has ensured that video content is being delivered to even the most basic devices and in limited network conditions. It has been one of the biggest enablers for video content delivery in APAC,” said Sood.
According to Murra, delivering video over IP infrastructure built for data transfer has many advantages. “Anyone plugged into a data network is a potential customer. IP leverages economies of scales with service providers potentially being able to reduce huge capex and opex.”
Due to the nature of data transfers over IP networks, however, buffering and latency are inherent to the way OTT is delivered and this massively impacts consumers’ QoE. There are multiple industry initiatives that are aimed at using a variety of techniques to reduce the latency and buffering.
The idea behind HEVC (H.265) is to provide the same quality of picture as its predecessor AVC (H.264) but with better compression so that there’s less data to deal with. Added Sood, “Instead of encoding every pixel from every frame in a video, HEVC encodes a single initial frame and then identifies the changes in each subsequent frame and encodes only the change.”
“While direct encoding and encapsulation into IP may be commonplace in cameras, a great deal of content is still SDI derived. Most of the content requires SDI conversion and encapsulation into IP prior to encoding,” Trow emphasised.
ABR Pros & Cons
Adaptive bit rate (ABR) streaming technologies have been successful in faithfully delivering content to users based on the network bandwidth experienced on their devices. When the connection is good, the viewer gets a high-quality, high-data-rate stream. If the connection speed drops, the server will send a lower-data-rate file to ensure a continuous connection, albeit at lower quality.
“The cons of ABR are more related to backend, especially at CDNs as they have to manage billions of segmented objects and block sizes of their storage solutions thereby leading to incremental overheads. But with economies of scale, this cost overhead will be easily offset leading to significantly scalable delivery solutions,” Sood pointed out.
Trow, meanwhile, said that ABR techniques provided the vital link between client CPU resource and the available delivery profile. “By encoding a range of profiles, a client can adaptively select the profile that best meets the capability of the broadband network.”
Currently, there are four main ABR formats: HTTP Live streaming (HLS), HTTP Smooth Streaming (HSS), HTTP Dynamic Streaming (HDS), and MPEG-DASH that has different versions. “The most popular format is HLS, which addresses Apple devices. However, MPEG-DASH adoption is growing and there are a few providers that use HSS. The main benefit of ABR streaming is that it adapts the quality of the video to the bandwidth available. Several layers are created, corresponding to a different bitrate, for each video,” Nouvel said. The entire process is based on a simple protocol–HTTP–that does not require heavy processing for delivery.
“The main drawback of ABR streaming is that it usually requires some buffering at the player level, inducing latency for live content. It also brings scalability issues when associated to unicast delivery even for live content. These drawbacks can be countered by using multicast ABR technologies,” she added.
Depending upon the requirement, there are different ways to publish ABR content.
“One way is HTTP-based approaches such as HLS and HSS wherein the player is responsible for requesting different streams according to CPU utilisation while the second is the server-based system, like RTMP, in which the streaming server is in charge of sending out the stream to the video player. More and more OTT players and CDNs are now moving towards adopting HTTP-based ABR due to the performance it offers,” Sood explained.
MPEG, Adobe, Apple, and Microsoft have all deployed ABR schemes. Trow added, “Both standardised and proprietary schemes exist for ABR handling. The variety of schemes that have been developed makes addressing the needs of multi-profile delivery a real challenge. Not only do the profiles have to deal with the broadband bit rate available, but they also need to contend with the subtly different ABR strategies to meet the needs of legacy players.” The aim of MPEG, in developing DASH, was to rationalise the complexity ABR deployments, at least in the short term. It represents another scheme to be implemented, admittedly with the long-term prospect of replacing legacy ABR strategies.
“CMAF will be supported by HLS and DASH and has the backing of big names such as Apple, Microsoft, and, most recently, Akamai. Legacy video-streaming players are going to be the biggest hurdle to universal adoption. Another issue with CMAF is that it is targeting HLS and DASH on the assumption that Smooth Streaming and HDS will be replaced by DASH.”
Publishing live video can be challenging and those actively engaged in providing channel provision are currently commissioning outsourced services at the CDN level.
Said Sood, “The encoded stream from a live event can be fetched and fed to any front-end, like a website or a mobile app, via streaming servers and a robust online video platform. Depending upon whether the live streaming is SD, HD, or 4K for encoding the live stream, new-age compression technologies like HEVC and AVC can be used.”
“The desire for many, particularly greenfield players, is to move to a service-based provision higher up the workflow but the requirements of live video and the reluctance of prominent broadcasters to relinquish control until they see a blueprint for such deployments are key issues,” added Trow.
Coming to compression, efficiency is the expectation of any implementation scheme. The advancements included within CMAF address live low-latency streaming applications.
Said Murra, “Today, there is really one codec used for video streaming—h.264—which has been used by the TV and media industry for 14 years. Its successor is a lot more unclear, perhaps, because there’s not going to be only one. HEVC has an unclear licensing system, which is proving a significant hurdle to mass market adoption; VP9 and, eventually, AV1 has been gaining ground through Netflix’ public endorsement.”
When it comes to live streaming, all codecs are not equal. “HEVC, which is mainly associated with Ultra HD, requires a lot more computing power at 3-5 times the capex and running costs of today’s codec and higher-cost decoding devices.” VP9 has found its niche in OTT today, although Google and Netflix are the only big players that have publicly announced their support for it. “Its costs of implementation are yet unknown and its higher computational requirements have put off companies that were already lukewarm about HEVC. Instead, what we are seeing is OTT players combining these various codecs for different use cases,” he added.
Today’s live streaming primarily relies on SD video but, as we prepare for UHD to become possible and HD to be commonplace, it is about consumers being able to access the best picture quality on their devices and networks.