IBC
article here
With audiences expecting instant global information, and broadcasters competing against social platforms that deliver clips within seconds, the race is on to deliver ultra-low latency (ULL) as a competitive edge.
“Latency and quality now go hand in hand,” explains Dr Anthony Basham, President, SCTE. “It’s great hearing the ball hit the net, but if you can’t see it because the picture is blown out, that’s no good.”
Although increasingly important to live sports, interactive formats, and real‑time streaming, low‑latency media delivery is most often impeded by the simplest, most integral steps in the content delivery workflow. For example, delays don’t only result from video processing, compression, and caching, but also from the redistribution of content over digital networks.
As the industry pushes to solve this crucial issue, IBC365 explores the two key, contrasting approaches to the ULL goal shaping media transport today.
The first is from BBC R&D, which emphasises open standards to achieve ULL at scale. The other, from Eluvio, claims to have demonstrated the same with a just-in-time software pipeline that runs as an application protocol on top of the Internet Protocol Suite (TCP/IP).
Eluvio targets satellite replacement
ULL streaming has long been one of the industry’s most persistent challenges, but Michelle Munson, CEO, Co‑Founder, Eluvio, believes the sector has finally reached a turning point. After a summer of large‑scale testing, she says the company’s Content Fabric has demonstrated that ULL can be delivered consistently, globally, and without the trade‑offs that have historically held back adoption.
During the FIFA World Cup, Eluvio streamed all 105 matches in a paid commercial pilot with a tier‑one US broadcaster, designed specifically to validate performance in a real‑world, direct‑to‑consumer environment. The results, Munson says, were unambiguous.
“We were ahead of: the broadcast by seven to 10 seconds per match; the best streaming service by 15 to 25 seconds; and the other streaming services by 30 seconds to a minute consistently.”
What impressed partners most, she says, was the uniformity of that performance. According to Munson, latency didn’t vary by region, and the system didn’t degrade under load. For global sports events – where millions watch the same moment simultaneously – that consistency is critical. Fans expect to see outcomes on the live stream first, not after social media or stadium cheers have already spoiled the moment.
However, the implications extend beyond consumer streaming. Eluvio’s stack also supports B2B broadcast feeds, which require extremely tight latency budgets. “Those have an imperative for ultra‑low latency in broadcast – well under 500 milliseconds, consistently around the world,” Munson reveals. “In the US, in some cases, we’re well under 20 milliseconds of broadcast.” That level of performance, she argues, is what enables realistic replacement of satellite distribution.
Munson rejects the idea that this is simply about gaining a small competitive edge. “When you’re 10 seconds ahead of broadcast, 15 seconds ahead of best‑case world streaming, and typically 30 seconds ahead, you’re in a different category,” she says. Historically, low‑latency solutions have forced compromises – sacrificing security, ease of deployment, universality, or cost. “Until now, it hasn’t really been something people could lean into because it was either compromising those other things… or it wasn’t good enough.”
Eluvio’s approach is based on the Content Fabric’s zero‑copy architecture. Instead of the traditional cloud/CDN workflow – transcoding, packaging, chunking, pushing through CDNs, and caching at the edge – the fabric creates a componentised, reference‑based representation of the content directly in the network. “It’s literally the same parts in the network,” Munson says. Live streams, VOD, clips and B2B feeds are all generated from the same underlying objects, without republishing or re‑packaging.
The benefits are substantial: no egress fees, no storage of multiple renditions, instant clip creation, and core bandwidth that remains constant regardless of audience size. This architecture also underpins the platform’s latency performance. The fabric’s just‑in‑time video processing pipeline fetches and assembles object parts with no added latency, delivering uniform results worldwide. Munson says full‑season customers see the same metrics “game after game after game.”
Compatibility is another pillar of the design. The fabric accepts standard broadcast inputs – secure reliable transport (SRT) protocol, transport stream, real-time messaging protocol (RTMP) – and outputs standards, dynamic adaptive streaming over HTTP (DASH) and HTTP Live Streaming (HLS). It supports high-efficiency video coding (HEVC) multiview, Widevine, FairPlay, Dolby Vision, Dolby Atmos, and HDR10. During the World Cup pilot, Munson says: “We were given the feeds literally a few days before it started and just go.” No workflow overhaul, no player changes, no custom APIs were needed.
The market response has been immediate. “Our company has just exploded this year,” Munson enthuses. Broadcasters want simplification, efficiency, and picture quality; leagues want global reach and new monetisation opportunities. Eluvio is now working with European Professional Club Rugby (EPCR), United Rugby Championship (URC), and Cricket Australia, all of whom have expanded their use of the platform. A major US tier‑one league – “second to none” – is also onboard, though not yet publicly named.
For Munson, the momentum reflects a broader shift in how live sports are consumed and valued. ULL is no longer a specialist feature – it’s becoming the baseline expectation for global events. Further, she believes the Content Fabric shows the industry doesn’t need to compromise to achieve it.
BBC edges closer to broadcast‑level streaming latency
As the UK edges toward an internet‑first broadcasting future, BBC R&D has been testing whether live IP streaming can finally match the speed and reliability of traditional broadcast. A large‑scale low‑latency DASH (LL‑DASH) trial, conducted across thousands of devices and viewer sessions during the Wimbledon Championships, suggests broadcast‑like delays are now achievable in real homes – and that the corporation may soon be ready to make the leap.
Phil Layton, Head, Applied Research, Media, BBC Research and Development, says the pressure from audiences has become impossible to ignore. “We always get viewer feedback about social media getting there first, or hearing cheers from next door,” he says. “People interact with our services differently now. The goal is to make IP as good or better than broadcast.”
That urgency has shaped the BBC’s low‑latency work since 2018, culminating in a three‑phase trial that tested LL‑DASH across more than 20,000 sessions and 5,000 IP addresses. The trial achieved a client latency target of 5.96 seconds and end‑to‑end delays of around nine seconds, taking into account encoding and transport – broadly comparable to digital terrestrial TV (DTT).
Chris Poole, Lead R&D Engineer, BBC, says the organisation’s decision to use LL‑DASH was driven by standards, compatibility, and cost. “LL‑DASH has advantages in backwards compatibility, which reduces deployment cost,” he rationalises. “We can encode the same media and make it available for low‑latency or normal delivery.”
Other emerging approaches, such as low latency, low delay (L3D) DASH, are on the BBC’s radar but were not tested within the trial’s timeframe. “We’re interested in how performance may differ,” Poole says.
The trial began with a small beta‑user test, then expanded to 21 device families – and their 500 model numbers – before scaling up during Wimbledon. The BBC provided a low‑latency stream of BBC Two’s coverage, reaching peak audiences of 30,000 and generating around 700,000 sessions.
Across all phases, the results were consistent: low‑latency streaming performed well, with only a slight increase in the proportion of viewers experiencing impairments such as buffering. “For every 100 viewers, low‑latency streaming added roughly one additional viewer experiencing more than a trivial impairment,” Poole says. “Performance was pretty good.”
Preparing for the future
The BBC believes that improvements to client‑side players can reduce some of those lingering impairments. “We see patterns suggesting the player could be improved,” Poole says. “With more work, performance could improve.”
Device variability remains a challenge. Layton says plug tests and test streams have not yet driven significant industry‑wide improvements. “Device manufacturers aren’t necessarily preparing for what’s coming,” he comments. The BBC has contributed enhancements to MPEG-DASH via JavaScript (dash.js) back to the open‑source community, but inconsistencies across TVs and streaming sticks remain. “If we deploy this, we want it in both iPlayer and Freely,” Layton adds. “There’s work to ensure all those devices are capable.”
The looming possibility of DTT switch‑off around 2034 adds pressure. Layton acknowledges the “ticking clock,” but stresses that the transition must be audience‑led. “By 2034, the number of DTT homes will be small,” he admits. “We want IP to be an attractive place for people to end up, not a forced transition.”
The BBC is also exploring future technologies, including Media over QUIC (MoQ). Poole says the corporation has long been interested in QUIC‑based video transport. “We’ll start with healthy scepticism,” he says. “We need to look at commercial deployment. It takes a long time to transition millions of devices.”
For now, the R&D team is reviewing the trial results and preparing recommendations. The decision to roll out low‑latency streaming across iPlayer will sit beyond the department. “We’re interested in other technologies as well,” Poole concludes. “We’re thinking about how we get more information on the pros and cons of those.”
What the trial has made clear is that broadcast‑like latency over IP is no longer theoretical. It is technically achievable, operationally viable, and – with further refinement – increasingly ready for prime time.
No comments:
Post a Comment