Ultra-Low Latency Streaming - Ant Media Server

Overview

Ultra-low latency streaming delivers live video with less than 1 second of delay, enabling truly interactive experiences. Ant Media Server achieves this through WebRTC technology, providing approximately 0.5 seconds of glass-to-glass latency.

What is Ultra-Low Latency?

Latency Comparison:

Protocol	Typical Latency	Use Case
WebRTC	0.5s	Real-time interaction
LL-HLS	2-6s	Low-latency at scale
LL-DASH	3-6s	Standards-based low latency
HLS	6-30s	Large-scale distribution
RTMP	3-5s	Legacy broadcasting

WebRTC Architecture

How WebRTC Achieves Low Latency

WebRTC eliminates traditional buffering through:

UDP Transport: No TCP retransmission delays
Minimal Buffering: Jitter buffer of ~50-200ms
Direct Peer Connection: No intermediate servers (in P2P mode)
Native Browser Support: No plugin overhead

WebRTC Components in Ant Media Server

The server implements WebRTC through several key interfaces:

IWebRTCAdaptor.java

public interface IWebRTCAdaptor {
    // Core WebRTC streaming interface
    // Handles peer connections, SDP negotiation, ICE
}

IWebRTCClient.java

public interface IWebRTCClient {
    // WebRTC client operations
    // Manages individual peer connections
}

IWebRTCMuxer.java

public interface IWebRTCMuxer {
    // WebRTC muxing operations
    // Handles media packaging for WebRTC delivery
}

Configuration

Basic WebRTC Settings

AppSettings.java:1001-1002

@Value ( "${webRTCEnabled:${" + SETTINGS_WEBRTC_ENABLED +":true}}" )
private boolean webRTCEnabled=true;

webRTCEnabled=true
webRTCFrameRate=30
webRTCKeyframeTime=2000

Port Configuration

WebRTC requires a range of UDP ports for media:

AppSettings.java:1307-1315

@Value( "${webRTCPortRangeMin:${" + SETTINGS_WEBRTC_PORT_RANGE_MIN +":50000}}")
private int webRTCPortRangeMin = 50000;

@Value( "${webRTCPortRangeMax:${" + SETTINGS_WEBRTC_PORT_RANGE_MAX +":60000}}")
private int webRTCPortRangeMax = 60000;

webRTCPortRangeMin=50000
webRTCPortRangeMax=60000

Ensure UDP ports 50000-60000 are open in your firewall for WebRTC to function properly.

ICE Configuration

ICE (Interactive Connectivity Establishment) handles NAT traversal:

AppSettings.java:1325-1340

@Value( "${stunServerURI:${" + SETTINGS_WEBRTC_STUN_SERVER_URI +":stun:stun1.l.google.com:19302}}")
private String stunServerURI = "stun:stun1.l.google.com:19302";

@Value( "${turnServerUsername:${" + SETTINGS_WEBRTC_TURN_SERVER_USERNAME +":}}")
private String turnServerUsername = "";

@Value( "${turnServerCredential:${" + SETTINGS_WEBRTC_TURN_SERVER_CREDENTIAL +":}}")
private String turnServerCredential = "";

STUN Configuration

stunServerURI=stun:stun1.l.google.com:19302

STUN servers help discover public IP addresses for NAT traversal.

TURN Configuration

For restrictive networks, configure TURN relay:

stunServerURI=turn:turn.example.com:3478
turnServerUsername=myuser
turnServerCredential=mypassword

TCP Candidates

Enable TCP fallback for restrictive firewalls:

AppSettings.java:1350

@Value( "${webRTCTcpCandidatesEnabled:${" + SETTINGS_WEBRTC_TCP_CANDIDATE_ENABLED +":false}}")
private boolean webRTCTcpCandidatesEnabled;

webRTCTcpCandidatesEnabled=true

TCP candidates increase latency slightly but improve connectivity in restricted networks.

SDP Semantics

AppSettings.java:1357-1358

@Value( "${webRTCSdpSemantics:${" + SETTINGS_WEBRTC_SDP_SEMANTICS +":" + SDP_SEMANTICS_UNIFIED_PLAN + "}}")
private String webRTCSdpSemantics = SDP_SEMANTICS_UNIFIED_PLAN;

webRTCSdpSemantics=unifiedPlan

Options:

unifiedPlan: Modern standard (recommended)
planB: Legacy format (deprecated)

Scaling WebRTC

Viewer Limits

Control concurrent WebRTC viewers per stream:

webRTCViewerLimit=-1

-1: Unlimited viewers
>0: Maximum concurrent WebRTC viewers per stream

SFU Mode vs. Transcoding

SFU (Selective Forwarding Unit)

When: No encoderSettingsString configured Benefits:

Zero transcoding overhead
Maximum server capacity
Minimal latency
CPU efficient

Limitations:

All viewers get same quality
No adaptive bitrate
Requires capable viewers

MCU (Multipoint Control Unit) with Transcoding

When: encoderSettingsString is configured Benefits:

Adaptive bitrate for viewers
Optimized for each device
Better viewer experience

Limitations:

Higher CPU usage
Slight latency increase (~100-200ms)
Resource intensive

Cluster Deployment

For high-scale WebRTC:

clusterCommunicationKey=yourSecretKey

Ant Media Server automatically:

Distributes load across nodes
Routes viewers to optimal servers
Handles failover

Performance Optimization

Codec Selection

Supported video codecs:

AppSettings.java:440-448

@Value( "${h264Enabled:${" + SETTINGS_H264_ENABLED +":true}}")
private boolean h264Enabled = true;

@Value( "${vp8Enabled:${" + SETTINGS_VP8_ENABLED +":true}}")
private boolean vp8Enabled = true;

@Value( "${h265Enabled:${" + SETTINGS_H265_ENABLED +":true}}")
private boolean h265Enabled = true;

h264Enabled=true
vp8Enabled=true
h265Enabled=false

Recommendations:

H.264: Best compatibility, good quality
VP8: Open-source, Chrome-optimized
H.265: Better compression, limited support

Frame Rate Control

AppSettings.java:1299-1300

@Value( "${webRTCFrameRate:${" + SETTINGS_WEBRTC_FRAME_RATE +":30}}" )
private int webRTCFrameRate = 30;

webRTCFrameRate=30

Lower frame rates reduce bandwidth:

30 fps: Standard quality
15 fps: Low bandwidth
60 fps: High motion content

Resolution & Bitrate Limits

maxResolutionAccept=1080
maxBitrateAccept=3000000
maxFpsAccept=30

Prevent resource abuse by limiting incoming streams.

Technical Deep Dive

Media Pipeline

WebRTC media flows through the muxer system:

Muxer.java:840-865

public synchronized boolean addVideoStream(int width, int height, AVRational timebase, int codecId, int streamIndex,
        boolean isAVC, AVCodecParameters codecpar) {
    boolean result = false;
    AVFormatContext outputContext = getOutputFormatContext();
    if (outputContext != null && isCodecSupported(codecId) && !isRunning.get())
    {
        registeredStreamIndexList.add(streamIndex);
        AVStream outStream = avformat_new_stream(outputContext, null);
        outStream.codecpar().width(width);
        outStream.codecpar().height(height);
        outStream.codecpar().codec_id(codecId);
        outStream.codecpar().codec_type(AVMEDIA_TYPE_VIDEO);
        // ... configuration continues
    }
}

Packet Processing

Minimal buffering for low latency:

Muxer.java:1204-1221

public synchronized void writePacket(AVPacket pkt, AVRational inputTimebase, AVRational outputTimebase, int codecType)
{
    AVFormatContext context = getOutputFormatContext();
    
    long pts = pkt.pts();
    long dts = pkt.dts();
    
    pkt.duration(av_rescale_q(pkt.duration(), inputTimebase, outputTimebase));
    pkt.pos(-1);
    
    totalSizeInBytes += pkt.size();
    
    currentTimeInSeconds = av_rescale_q(pkt.dts(), inputTimebase, avRationalTimeBase);
    // ... process and write packet
}

Use Cases

Live Auctions

Requirements:

Sub-second latency
Synchronized bidding
Mobile support

Configuration:

webRTCEnabled=true
webRTCFrameRate=30
statsBasedABREnabled=true

Video Conferencing

Requirements:

Bidirectional communication
Multiple participants
Screen sharing

Configuration:

webRTCEnabled=true
dataChannelEnabled=true
maxVideoTrackCount=9
maxAudioTrackCount=9

Interactive Gaming

Requirements:

Minimal latency
High frame rate
Low jitter

Configuration:

webRTCEnabled=true
webRTCFrameRate=60
abrUpScaleJitterMs=2

Surveillance

Requirements:

Real-time monitoring
Multiple cameras
Recording

Configuration:

webRTCEnabled=true
mp4MuxingEnabled=true
hlsMuxingEnabled=true

Trade-offs: WebRTC vs. HLS

When to Use WebRTC

Advantages:

Ultra-low latency (~0.5s)
Real-time interaction
Built-in encryption (DTLS/SRTP)
NAT traversal built-in

Limitations:

Higher server resources per viewer
Complex firewall configuration
Limited CDN support
Browser compatibility variations

When to Use HLS

Advantages:

Unlimited scalability via CDN
Universal device support
Lower server resources
Simple infrastructure

Limitations:

Higher latency (6-30s)
Not suitable for interaction
Adaptive bitrate complexity

Hybrid Approach

Ant Media Server can deliver both simultaneously:

webRTCEnabled=true
hlsMuxingEnabled=true

Use case: Interactive studio with mass audience

Studio participants use WebRTC
Mass audience uses HLS via CDN

Monitoring & Troubleshooting

WebRTC Statistics

Monitor real-time statistics:

curl "https://your-server:5443/AppName/rest/v2/broadcasts/{streamId}/webrtc/stats"

Response includes:

{
  "videoPacketsLost": 0,
  "audioPacketsLost": 0,
  "videoFramesDecoded": 1234,
  "audioLevel": 0.5,
  "videoRoundTripTime": 0.02,
  "videoJitter": 0.001
}

Common Issues

High Packet Loss

Symptoms: Choppy video, audio dropouts Solutions:

Enable stats-based ABR
Reduce bitrate
Check network quality
Configure TURN server

Connection Failures

Symptoms: Unable to establish connection Solutions:

Verify STUN/TURN configuration
Check firewall UDP ports
Enable TCP candidates
Test with different networks

High Latency

Symptoms: >2 seconds delay Solutions:

Disable transcoding (use SFU mode)
Reduce jitter buffer
Check server resources
Optimize network routing

Next Steps

Streaming Protocols

Compare WebRTC with other protocols

Architecture

Understand WebRTC in the system architecture

​Overview

​What is Ultra-Low Latency?

​WebRTC Architecture

​How WebRTC Achieves Low Latency

​WebRTC Components in Ant Media Server

​Configuration

​Basic WebRTC Settings

​Port Configuration

​ICE Configuration

​STUN Configuration

​TURN Configuration

​TCP Candidates

​SDP Semantics

​Scaling WebRTC

​Viewer Limits

​SFU Mode vs. Transcoding

​SFU (Selective Forwarding Unit)

​MCU (Multipoint Control Unit) with Transcoding

​Cluster Deployment

​Performance Optimization

​Codec Selection

​Frame Rate Control

​Resolution & Bitrate Limits

​Technical Deep Dive

​Media Pipeline

​Packet Processing

​Use Cases

​Live Auctions

​Video Conferencing

​Interactive Gaming

​Surveillance

​Trade-offs: WebRTC vs. HLS

​When to Use WebRTC

​When to Use HLS

​Hybrid Approach

​Monitoring & Troubleshooting

​WebRTC Statistics

​Common Issues

​High Packet Loss

​Connection Failures

​High Latency

​Next Steps

Streaming Protocols

Architecture

Overview

What is Ultra-Low Latency?

WebRTC Architecture

How WebRTC Achieves Low Latency

WebRTC Components in Ant Media Server

Configuration

Basic WebRTC Settings

Port Configuration

ICE Configuration

STUN Configuration

TURN Configuration

TCP Candidates

SDP Semantics

Scaling WebRTC

Viewer Limits

SFU Mode vs. Transcoding

SFU (Selective Forwarding Unit)

MCU (Multipoint Control Unit) with Transcoding

Cluster Deployment

Performance Optimization

Codec Selection

Frame Rate Control

Resolution & Bitrate Limits

Technical Deep Dive

Media Pipeline

Packet Processing

Use Cases

Live Auctions

Video Conferencing

Interactive Gaming

Surveillance

Trade-offs: WebRTC vs. HLS

When to Use WebRTC

When to Use HLS

Hybrid Approach

Monitoring & Troubleshooting

WebRTC Statistics

Common Issues

High Packet Loss

Connection Failures

High Latency

Next Steps