Live streaming has become extremely popular in recent years, allowing people to broadcast events in real time to viewers across the globe. However, one common frustration with live streaming is delay – the lag between when something happens on camera and when viewers see it on their screens. This delay can range from a few seconds to over a minute, which can be very disruptive for live events. So how can you reduce or eliminate delay when live streaming?
What Causes Delay in Live Streaming?
There are a few main culprits behind live streaming delay:
- Encoding – The camera feed needs to be digitized and encoded into a video format like H.264. This encoding process takes time.
- Packeting – The encoded video is broken up into many small network packets for transmission. Packetizing the data also adds a bit of delay.
- Buffering – Packets may arrive at different times or get lost, so the player has to buffer a few seconds worth of video to account for network fluctuations.
- Processing – The player has to receive the packets, reassemble them, decode the video, and then display it. More software processing equals more delay.
In general, the more steps in the live streaming workflow, the more potential points where latency can creep in. The total end-to-end delay is the sum of small delays introduced at each stage.
Reducing Encoding Time
The encoding process causes a significant chunk of live streaming latency. Using a hardware encoder rather than software encoding can help speed up the encoding process. Hardware encoders leverage specialized chips like GPUs or FPGAs to encode video much faster than software running on a generic CPU.
Choosing an efficient codec like H.264 or H.265 over an older codec like MPEG-2 will also result in faster encoding times and thus lower latency. The newer H.26x codec family uses smarter encoding techniques to compress video into a smaller file size versus older codecs.
Lowering your video resolution and frame rate can also result in faster encoding. Pushing fewer pixels means less data to encode. Just be mindful that too low of resolution or frame rate may impact quality.
Using a Low-Latency Streaming Protocol
Standard streaming protocols like RTMP have latency baked into their designs. RTMP has a delay of 5+ seconds since it needs to fill TCP buffers and retransmit dropped packets.
Switching to a protocol designed for low-latency live streaming can greatly reduce delay:
- WebRTC – Leverages UDP with built-in packet loss resilience. Sub-second latency possible.
- SRT – Uses UDP while minimizing packet loss through its own error correction. Latency under 1 second.
- RTMPS – An optimized version of RTMP using Secure TCP. Can get latency around 1-2 seconds.
The downside is these protocols may require custom players rather than mainstream ones like Flash or HTML5 video tag. But for live streaming where low delay is critical, using one of these protocols is recommended.
Optimizing Your CDN
Using a content delivery network (CDN) involves adding more network hops which can increase latency. Optimize your CDN setup to minimize impact on delay:
- Ensure your CDN has edge servers located geographically close to your viewers.
- Configure your CDN for low-latency delivery or live streaming if available.
- Use CDN chaining to pass streams through fewer CDN networks.
- Fine-tune edge server caching rules to cache as little as possible.
Monitoring CDN performance and strategically picking edge server locations are key to reducing CDN latency impact.
Player Buffering and Optimization
The player is responsible for the final lag as it must buffer packets before decoding and playback. To reduce player delay:
- Use a player designed for low-latency playback. Test different players.
- Reduce player buffering time to the bare minimum to avoid freezing.
- Disable pre-decoding or decoding ahead for smoother playback.
- Optimize player decoding – leverage GPU, multithreading.
Balance buffering versus latency when tuning the player. Remember that a faster live stream is pointless if it keeps freezing and rebuffering.
Reducing Overall Pipeline Latency
In addition to optimizing individual components, also evaluate your complete streaming pipeline end-to-end. Here are some holistic tips for reducing total latency:
- Simplify your workflow – eliminate unnecessary steps between camera and viewer.
- Use modern low-latency technologies – encoder, protocol, CDN, player.
- Tune latency vs. quality – higher latency enables better compression and quality.
- Monitor delay at each step to identify problem points.
- Analyze end-user metrics – playback freezes, rebuffers, lag behind live.
Benchmark different setups under real viewer load to validate overall performance. Always keep the end user experience in mind when optimizing latency.
When Is Low Latency Streaming Important?
Here are some examples where having the lowest possible live streaming latency is crucial:
- Live sports – Fans want to experience goals, touchdowns, etc. as they happen live.
- Esports gaming – Gamers need to see gameplay with minimal delay.
- Live betting – Bettors need real-time info to place in-game bets.
- Finance – Traders need fast stock price updates to make transactions.
- Online auctions – Bidders need to see current bid price with little lag.
High interactivity and real-time feedback are common requirements for use cases demanding low streaming latency. Entertainment, sports, and finance are popular genres.
When Higher Latency Is Acceptable
For some types of streaming content, having some delay is perfectly acceptable:
- Corporate communications – executive messaging, town halls, presentations
- Concerts – fans enjoy the music and performances
- News broadcasts – getting info slightly delayed is fine
- Classroom lectures – students still learn from past lessons
- Religious services – worshippers still get the message
Here interactivity is not as essential. Viewers are more tolerant of higher latency as long as the content eventually arrives. Video quality and stability matter more than pure low latency.
Typical Live Streaming Latency Values
To set expectations, here are rough latency ranges for different live streaming configurations:
| Configuration | Typical Latency |
| Hardware encoder + SRT | 0.5 – 2 seconds |
| Hardware encoder + WebRTC | 1 – 3 seconds |
| Software encoder + RTMP | 5 – 10 seconds |
| Cloud transcoding + CDN | 10 – 30 seconds |
Actual values depend on many factors – codec, protocols, buffering, etc. But this table gives a general idea for delay ranges.
How Low Can You Go?
Is there a physical limit to how low live streaming latency can be? Light in vacuum travels at about 300,000 km/sec. So for a viewer watching a stream hosted 500 km away, the propagation delay is:
Distance / Speed of Light = 500 km / 300,000 km/sec = 0.0017 sec = 1.7 ms
So while we can’t break the laws of physics, we can approach streaming latency of just a few milliseconds with today’s technology and global infrastructure. Extremely low latency streaming will continue improving as newer technologies emerge.
Conclusion
Reducing live streaming delay involves optimizing every step of the pipeline – encoding, transport, buffering, and playback. Choosing low-latency components like modern codecs and protocols gives you a head start. Tuning latency versus stability and quality is also key. With proper design and testing, you can achieve sub-second streaming latency suitable for live sports, esports, betting, finance, and other interactive use cases.