Implementing WebRTC Video with Jitter Buffer and NACK in a Qt Client
WebRTC uses UDP to minimize transmission overhead. Clients exchange SDP descriptors, use STUN/TURN for NAT traversal, ICE to maintain connections, and a signaling server for coordination. SFU/MCU servers act as middleboxes for stream forwarding. A 1080p@30fps VP8 stream requires ~600 Kbps without key frames. Packet loss dramatically increases load through PLI requests and NACK. The jitter buffer resolves packet ordering issues.
Key tasks: signaling/MCU server with NACK support, retransmission mechanism, client-side jitter buffer.
Pion Server with NACK Support
The server is implemented in Go using the Pion library. We register default interceptors for NACK:
type Controller interface {
HandleConnection(c *common.SafeWebSocket)
JoinRoom(peer *common.Peer, msg Msg) error
LeaveRoom(peer *common.Peer, msg Msg) error
}
type controller struct {
logger *zap.Logger
roomRepo repository.RoomRepo
api *webrtc.API
}
func NewController(logger *zap.Logger, roomRepo repository.RoomRepo) Controller {
settingEngine := webrtc.SettingEngine{}
settingEngine.SetAnsweringDTLSRole(webrtc.DTLSRoleServer)
mediaEngine := &webrtc.MediaEngine{}
mediaEngine.RegisterDefaultCodecs()
interceptorRegistry := interceptor.Registry{}
if err := webrtc.RegisterDefaultInterceptorsWithOptions(mediaEngine, &interceptorRegistry,
webrtc.WithNackGeneratorOptions(nack.GeneratorSize(8192)),
webrtc.WithNackResponderOptions(nack.ResponderSize(8192)),
); err != nil {
logger.Error("failed to register interceptor", zap.Error(err))
panic(err)
}
api := webrtc.NewAPI(
webrtc.WithMediaEngine(mediaEngine),
webrtc.WithSettingEngine(settingEngine),
webrtc.WithInterceptorRegistry(&interceptorRegistry),
)
ctrl := &controller{
api: api,
logger: logger,
roomRepo: roomRepo,
}
go func() { // Send RTCP I-frame request every two seconds
ticker := time.NewTicker(2 * time.Second)
for _ := range ticker.C {
roomIds := ctrl.roomRepo.GetRooms()
for _, roomId := range roomIds {
go ctrl.dispatch(roomId)
}
}
}()
return ctrl
}
Network emulation with packet loss:
sudo tc qdisc add dev lo root netem delay 50ms 20ms loss 1%
Qt Client Using libdatachannel
The browser serves as the video source via a minimal HTTP server. The C++ client uses libdatachannel + Qt without requiring full Chromium.
Establishing peer connection:
void ConferenceClient::connectClient(QString url, QString roomId)
{
rtc::InitLogger(rtc::LogLevel::Debug);
this->pc.onLocalDescription(this->pcOnLocalDescription(roomId));
this->pc.onLocalCandidate(this->pcOnLocalCandidate());
this->pc.onGatheringStateChange(this->pcOnGatheringStateChange());
this->pc.onIceStateChange([](auto state) {
std::cout << "Ice state changed: " << state << std::endl;
});
this->pc.onStateChange([](auto state) {
std::cout << "state changed: " << state << std::endl;
});
this->ws.onOpen(this->wsOnOpen(roomId));
this->ws.onMessage(this->wsOnMessage());
this->pc.onTrack(this->pcOnTrack());
this->ws.open(url.toStdString());
}
Handling Tracks and Packets
For each track, we create a structure with a jitter buffer (LRUCache) and a frame queue:
std::function<void(std::shared_ptr<rtc::Track>)> ConferenceClient::pcOnTrack() {
return [this](std::shared_ptr<rtc::Track> track) {
auto mid = track->description().mid();
this->track_index[mid]
= {track, 0, "NO_VALUE", 0, 0, LRUCache<std::uint32_t, jitterbuffer>(256)};
this->player->initMid(mid);
bool isVideo = true;
if (track->description().type() == "audio") {
isVideo = false;
track->setMediaHandler(std::make_shared<rtc::OpusRtpDepacketizer>());
track->chainMediaHandler(std::make_shared<rtc::RtcpReceivingSession>());
track->onFrame(this->trackOnFrame(mid, isVideo));
} else {
track->onMessage(this->pcOnMessage(mid));
}
track->onOpen([track]() { track->requestKeyframe(); });
track->onClosed([this, mid]() { this->player->destroy(mid); });
};
}
Key aspects of RTP packet handling:
- Discarding late packets (rtpHeader->timestamp() < lastCompletedTs)
- RTX retransmission – restoring original seqNumber and payloadType
- Endianness – RTP data is big-endian; platform is little-endian
- Frame queue as std::map<uint32_t, pair<long, vector<byte>>> indexed by RTP timestamp
- Playback after PLAYER_DELAY following first frame packet arrival
Reassembling VP8/VP9 Frames
std::function<void(rtc::message_variant)> ConferenceClient::pcOnMessage(std::string mid)
{
return [this, mid](rtc::message_variant msg) {
// ... timestamp check, RTX processing ...
std::vector<std::byte> frame;
if (!frame_cache.exist(pkgTs)
&& (track->description().rtpMap(PT)->format == MyApp::VP8CODEC
|| track->description().rtpMap(PT)->format == MyApp::VP9CODEC)) {
frame_cache.put(pkgTs, jitterbuffer());
track_info.ssrc = rtpHeader->ssrc();
track_info.frame_queue[pkgTs] = std::make_pair(nowTs, std::vector<std::byte>());
codec = track->description().rtpMap(PT)->format;
codecPT = PT;
}
jitterbuffer &buff = frame_cache.get(pkgTs);
if (codec == MyApp::VP9CODEC) {
frame = buff.addVp9Packet(std::move(msg), track_info.lastCompletedTs);
} else if (codec == MyApp::VP8CODEC) {
frame = buff.addVp8Packet(std::move(msg), track_info.lastCompletedTs);
}
if (frame.size() > 0) {
track_info.frame_queue[pkgTs].second = std::move(frame);
}
// ... playback and NACK ...
};
}
Key Takeaways
- libdatachannel requires manual implementation of jitter buffer and NACK
- Pion interceptors automatically handle NACK with buffer sizes set to 8192
- Frame queue uses std::map for natural ordering by RTP timestamp
- RTX mappings are negotiated in SDP offer/answer
- Playback occurs after PLAYER_DELAY post-first-frame arrival
- Periodic RTCP PLI sent every 2 seconds
— Editorial Team
No comments yet.