HQ Trivia and the Race to Ultra-Low Latency Live Streaming

Mobile gaming sensation HQ Trivia is being hyped as the future of mobile video and of live TV. Indeed, that is what its makers, Intermedia Labs, have suggested.

HQ Trivia is a free, interactive iPhone game (with Android in beta) that mimics the format of a TV quiz show. Multiple-choice quizzes are run twice a day with a typical pool of $10,000 split between hundreds of winners.

Aside from its cash enticement, many see the real success of the app as appointment gaming—its ability to drive viewers to a "TV-like" schedule on mobile.

Despite the game's popularity—more than 1 million play daily, and that number continues to rise, with separate versions for the U.S. and UK—there are still questions about its longevity, not least its ability to scale and to solve glitches in latency, both of which are essential for the growth of the live massive multiplayer video experience.

Low-latency live streaming is the only way to get the true "live" experience making it possible, in the case of HQ Trivia, for the host to quickly respond to results, chats and users.

According to Geert Faber VP, Americas at multiscreen strategists Ex Machina Group, HQ Trivia is attempting to combine two challenging technologies in a single experience: low-latency live streaming and the scaling up of the trivia gameplay.

"[It] needs to send the questions, register the user interaction, and provide the correct response. In addition, it needs to show if a user got it right and how many players are left, all without providing questions and answers beforehand," he says. "Processing hundreds of thousands of interactions in split seconds and calculating results in real time can be a load balance and server challenge."

Intermedia Labs also recently added a social function in which players can now see which friends and family members are playing at the same time.

The challenge with scheduled live events is meeting the extreme peak in traffic, usually when the program starts, without overloading the network and servers. HQ Trivia has acknowledged issues on its Twitter feed: "We’re working on making the service more reliable as we scale to meet demand," said a message from November.

While Intermedia Labs have been coy about unveiling the streaming tech that lies under the hood, there are reports that it is using a combination of RTMP and RTSP and a proprietary video player.

RTMP is of course a proven delivery technology, but one that’s also one on the wane as more and more CDNs end their support for Flash streaming. On the other hand, other HTTP-based protocols like HLS and MPEG-DASH, while supported by all the majority of CDNs and players, are optimised for broadcast, not real-time interactivity.

Another option might be WebRTC, the real-time communication technology promoted by Google. This was designed for peer-to-peer communication in small groups only, and it doesn’t scale well for large number of viewers. Plus, it is missing CDN and vendor support (it doesn’t work on Safari).

While HQ Trivia has succeeded in showing the appeal of an interactive and social broadcast direct to people’s phones, it won’t be the only such app for long.

Video is turning away from broadcast toward two-way or multi-way communication, and webcasters and broadcasters want to be able to communicate directly in real time or engage with their audiences by getting live feedback during presentations or events. Livestreamed gaming, auction, betting, and video chat, depend on latency below 1 or 2 seconds for the experience.

"Flash is dying, and HLS and DASH cannot deliver the same performance for live streams," says Oliver Lietz, founder at video encoding and streaming software developer nanocosmos. "Latencies of H.264 GOP is usually of two seconds or more, overall resulting in 10-30 seconds or more due to additional segmentation. Additionally, latency will increase for network dropouts and instabilities. Interactive applications cannot work under these conditions."

RTMP is still a valid and suitable technology, Lietz says, for low-latency high-quality live video transmission, but a more "intelligent" delivery format is needed to keep latency low and independent from constraints like GOP lengths.

"Live streaming interactivity requires a bidirectional connection with ultra-low latency (below one or two seconds), to achieve real time user experiences in environments like live auctions, gambling, webcasts, or sports," he says.

Ultra-Low Latency

Live streaming platform developer Wowzasees the same trend. "From HQ with live trivia and interactive gaming, to ecommerce platforms like Mercari who are implementing live streaming to power live auctions, more applications are driven by audience interaction," says director of communications Chris Michaels. "What ties them together? Ultra-low latency delivery and interactivity.

"Most of these apps are doing more than video—they’re providing real-time data or interactive layers with the video," he adds. "That requires two-way communication between the server and client, which in a traditional TCP infrastructure could easily bloat bandwidth requirements with server chatter."

The race is on to develop and prove Ultra-Low Latency (ULL) services.

Ex Machina’s answer is its PlayToTV platform that it claims can handle 1 million concurrent users with the capacity to scale from zero to 400,000 users "in a split second" for real-time user interaction.

Wowza offers an Ultra-Low Latency Service as part of the Wowza Streaming Cloud for global delivery in 2 seconds or less, it claims.

The biggest challenge for any of the ULL delivery is playback. Wowza’s solution uses the WOWZ protocol and connects via Websockets to deliver to both the Wowza Player (an HTML5-embeddable web player) as well as its Wowza GoCoder SDK for mobile playback on iOS and Android devices.

"We specifically built our ULL solution on the WOWZ protocol, as it works as an RTMP-like solution and doesn’t suffer from the scalability or playback problems that WebRTC architectures experience today; or the security concerns with peering and UDP," explains Michaels.

That said, Wowza does offer a number of options. It supports WebRTC, SRT for ingest and midgress, and RTMP/RTSP workflows. "We’re always investigating new technologies like LHLS, CMAF and others that may provide viable solutions for our customers," he says.

However, Wowza’s ULL is for "in private availability" only. "Not everyone can access it today through the Wowza Streaming Cloud manager, but they can always contact us for pricing and availability," says Michaels. "We have had a number of customers from live sports, auctions and gaming, to real-time chat (1:many, and many:many scenarios) that have had a lot of success using the platform."

The Open Source Solution

Then there is SRT (Secure Reliable Transport) developed by Haivisionand made open-source (non-proprietary)in April 2017, subsequently endorsedby over 50 companies spanning internet streaming (Wowza, Limelight, etc.) and broadcast contribution (Harmonic, Eurovision, etc.) markets and applications.

It iscurrently used intwo distinct use cases: Streamers who want to solve the first-milechallenges associated with RTMP; and broadcasters requiring low latency contribution and who want to mitigate the costs ofprivate fibre networks or satellite. "To satisfy these challenges, SRTis a very compelling alternative to costly proprietary solutions," says Haivision CMO Peter Maag. "SRT far surpasses RTMP with regard tolatency, security, and bandwidth utilization.

The NFL, NBA, and the Press Association/Globelynx are using SRT for interactive sportsinterviews, he says. SRT is also fueling daily, event-based, 24/7 broadcasting forportals like FuboTV, Tencent, and Cheddar.

However, SRT is currently designed for one-to-one or one-to-few (up to 1000) endpoints. According to Maag, "SRT may extend to be a solution to address one to many but it’s not there yet. The market needs to drive towards standards-based, broad scale, low latency delivery. Frameworks like Google QUIC are vying to minimize delivery to player latency for mass audiences.

Maag suggests that HQ Trivia’scombination of synchronization and acceptable latency
(which he estimates at less than 12 seconds round trip) delivery at scale is "amazing at drivingengagement through interactivity.

"Where HQ Trivia pulls this off at 12 seconds oflatency, the market is striving for less than 5 seconds at scale orlower," he says.

Building on Standards

According to Lietz, "CDN vendors either do not care about latency or try to reduce latency down to around 6-10 seconds."

Nanocosmos’ own target latency for its nanoStream H5Live is "around 1 second" end-to-end for full interactive live streaming user experiences.

Nanocosmos is positioning nanoStream H5Live as a new ULL playback technology standard which takes existing standards (WebSocket, HLS, fMP4) and builds them into a HTML5 playback solution—including Safari on iOS—for any device, plugin-free on all browsers.

"The industry is entrenched in its way of thinking about HLS, DASH and WebRTC," says Lietz. "We embrace these technologies and have added something unique to them to produce genuine ULL delivery. We’ve not reinvented the wheel with a new standard, we’ve enhanced existing ones."

He adds, "As our server component is very lightweight, our cloud service can scale well for thousands of viewers. We are working with CDN partners for unlimited numbers of viewers. There are no general technical limits."

Regardless, inorder to achieve ubiquity, the key to success for ULL is nativeadoption across all mobile devices and browsers, a challenge that willbe ‘arm wrestled’ only by the top vendors in the market.

"Scaling is going to be tricky for any IP-delivered solution, as you have to either accurately guess the peak viewers, or have an elastic solution that can handle the thundering herd," says Michaels. "And, if you consider low-latency delivery, you’ve added a factor of 10x for possible problems. There’s no one cure-all today to fix the problem. It’s so multifaceted."