feat: protocol improvements — live trunking, mini-frames, noise suppression, adaptive jitter

T6 wiring: Trunking in relay hot path
- TrunkedForwarder wraps transport with TrunkBatcher
- run_participant uses 5ms flush timer when trunking enabled
- send_trunk/recv_trunk on QuinnTransport
- --trunking flag on relay config
- 2 new tests: forwarder batches, auto-flush on full

T7 wiring: Mini-frames in encoder/decoder
- MediaPacket::encode_compact/decode_compact with MiniFrameContext
- CallEncoder sends mini-headers for consecutive frames (full every 50th)
- CallDecoder auto-detects full vs mini on receive
- mini_frames_enabled in CallConfig (default true)
- 3 new tests: encode/decode sequence, periodic full, disabled mode

Noise suppression (nnnoiseless/RNNoise)
- NoiseSupressor in wzp-codec: pure Rust ML-based noise removal
- Processes 960-sample frames as two 480-sample halves
- Integrated in CallEncoder before silence detection
- noise_suppression in CallConfig (default true)
- 4 new tests: creation, processing, SNR improvement, passthrough

T1-S4: Adaptive playout delay
- AdaptivePlayoutDelay: EMA-based jitter tracking (NetEq-inspired)
- Computes target_delay from observed inter-arrival jitter
- JitterBuffer::new_adaptive() uses adaptive delay
- adaptive_jitter in CallConfig (default true)
- 5 new tests: stable, jitter increase, recovery, clamping, estimate

272 tests passing across all crates.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Siavash Sameni
2026-03-28 14:24:53 +04:00
parent 34cd1017c1
commit 0dc381e948
11 changed files with 1547 additions and 11 deletions

View File

@@ -33,6 +33,12 @@ pub struct RelayConfig {
/// Discovery is manual via multiple --probe flags; this flag signals intent.
#[serde(default)]
pub probe_mesh: bool,
/// Enable trunk batching for outgoing media in room mode.
/// When true, packets destined for the same receiver are accumulated into
/// [`TrunkFrame`]s and flushed every 5 ms (or when the batcher is full),
/// reducing per-packet QUIC datagram overhead.
#[serde(default)]
pub trunking_enabled: bool,
}
impl Default for RelayConfig {
@@ -48,6 +54,7 @@ impl Default for RelayConfig {
metrics_port: None,
probe_targets: Vec::new(),
probe_mesh: false,
trunking_enabled: false,
}
}
}

View File

@@ -64,6 +64,9 @@ fn parse_args() -> RelayConfig {
"--probe-mesh" => {
config.probe_mesh = true;
}
"--trunking" => {
config.trunking_enabled = true;
}
"--mesh-status" => {
// Print mesh table from a fresh registry and exit.
// In practice this is useful after the relay has been running;
@@ -84,6 +87,7 @@ fn parse_args() -> RelayConfig {
eprintln!(" --probe <addr> Peer relay to probe for health monitoring (repeatable).");
eprintln!(" --probe-mesh Enable mesh mode (mark config flag, probes all --probe targets).");
eprintln!(" --mesh-status Print mesh health table and exit (diagnostic).");
eprintln!(" --trunking Enable trunk batching for outgoing media in room mode.");
eprintln!();
eprintln!("Room mode (default):");
eprintln!(" Clients join rooms by name. Packets forwarded to all others (SFU).");
@@ -239,6 +243,7 @@ async fn main() -> anyhow::Result<()> {
let auth_url = config.auth_url.clone();
let relay_seed_bytes = relay_seed.0;
let metrics = metrics.clone();
let trunking_enabled = config.trunking_enabled;
tokio::spawn(async move {
let addr = connection.remote_address();
@@ -423,6 +428,7 @@ async fn main() -> anyhow::Result<()> {
transport.clone(),
metrics.clone(),
&session_id_str,
trunking_enabled,
).await;
// Participant disconnected — clean up per-session metrics

View File

@@ -6,13 +6,17 @@
use std::collections::{HashMap, HashSet};
use std::sync::atomic::{AtomicU64, Ordering};
use std::sync::Arc;
use std::time::Duration;
use bytes::Bytes;
use tokio::sync::Mutex;
use tracing::{error, info, warn};
use wzp_proto::packet::TrunkFrame;
use wzp_proto::MediaTransport;
use crate::metrics::RelayMetrics;
use crate::trunk::TrunkBatcher;
/// Unique participant ID within a room.
pub type ParticipantId = u64;
@@ -171,8 +175,70 @@ impl RoomManager {
}
}
// ---------------------------------------------------------------------------
// TrunkedForwarder — wraps a transport and batches outgoing media into trunk
// frames so multiple packets ride a single QUIC datagram.
// ---------------------------------------------------------------------------
/// Wraps a [`QuinnTransport`] with a [`TrunkBatcher`] so that small media
/// packets are accumulated and sent together in a single QUIC datagram.
pub struct TrunkedForwarder {
transport: Arc<wzp_transport::QuinnTransport>,
batcher: TrunkBatcher,
session_id: [u8; 2],
}
impl TrunkedForwarder {
/// Create a new trunked forwarder.
///
/// `session_id` tags every entry pushed into the batcher so the receiver
/// can demultiplex packets by session.
pub fn new(transport: Arc<wzp_transport::QuinnTransport>, session_id: [u8; 2]) -> Self {
Self {
transport,
batcher: TrunkBatcher::new(),
session_id,
}
}
/// Push a media packet into the batcher. If the batcher is full it will
/// flush automatically and the resulting trunk frame is sent immediately.
pub async fn send(&mut self, pkt: &wzp_proto::MediaPacket) -> anyhow::Result<()> {
let payload: Bytes = pkt.to_bytes();
if let Some(frame) = self.batcher.push(self.session_id, payload) {
self.send_frame(&frame)?;
}
Ok(())
}
/// Flush any pending packets — called on the 5 ms timer tick.
pub async fn flush(&mut self) -> anyhow::Result<()> {
if let Some(frame) = self.batcher.flush() {
self.send_frame(&frame)?;
}
Ok(())
}
/// Return the flush interval configured on the inner batcher.
pub fn flush_interval(&self) -> Duration {
self.batcher.flush_interval
}
fn send_frame(&self, frame: &TrunkFrame) -> anyhow::Result<()> {
self.transport.send_trunk(frame).map_err(|e| anyhow::anyhow!(e))
}
}
// ---------------------------------------------------------------------------
// run_participant — the hot-path forwarding loop
// ---------------------------------------------------------------------------
/// Run the receive loop for one participant in a room.
/// Forwards all received packets to every other participant.
///
/// When `trunking_enabled` is true, outgoing packets are accumulated per-peer
/// into [`TrunkedForwarder`]s and flushed every 5 ms or when the batcher is
/// full, reducing QUIC datagram overhead.
pub async fn run_participant(
room_mgr: Arc<Mutex<RoomManager>>,
room_name: String,
@@ -180,6 +246,29 @@ pub async fn run_participant(
transport: Arc<wzp_transport::QuinnTransport>,
metrics: Arc<RelayMetrics>,
session_id: &str,
trunking_enabled: bool,
) {
if trunking_enabled {
run_participant_trunked(
room_mgr, room_name, participant_id, transport, metrics, session_id,
)
.await;
} else {
run_participant_plain(
room_mgr, room_name, participant_id, transport, metrics, session_id,
)
.await;
}
}
/// Plain (non-trunked) forwarding loop — original behaviour.
async fn run_participant_plain(
room_mgr: Arc<Mutex<RoomManager>>,
room_name: String,
participant_id: ParticipantId,
transport: Arc<wzp_transport::QuinnTransport>,
metrics: Arc<RelayMetrics>,
session_id: &str,
) {
let addr = transport.connection().remote_address();
let mut packets_forwarded = 0u64;
@@ -242,6 +331,120 @@ pub async fn run_participant(
mgr.leave(&room_name, participant_id);
}
/// Trunked forwarding loop — batches outgoing packets per peer.
async fn run_participant_trunked(
room_mgr: Arc<Mutex<RoomManager>>,
room_name: String,
participant_id: ParticipantId,
transport: Arc<wzp_transport::QuinnTransport>,
metrics: Arc<RelayMetrics>,
session_id: &str,
) {
use std::collections::HashMap;
let addr = transport.connection().remote_address();
let mut packets_forwarded = 0u64;
// Per-peer TrunkedForwarders, keyed by the raw pointer of the peer
// transport (stable for the Arc's lifetime). We use the remote address
// string as the key since it is unique per connection.
let mut forwarders: HashMap<std::net::SocketAddr, TrunkedForwarder> = HashMap::new();
// Derive a 2-byte session tag from the session_id hex string.
let sid_bytes: [u8; 2] = parse_session_id_bytes(session_id);
let mut flush_interval = tokio::time::interval(Duration::from_millis(5));
// Don't let missed ticks pile up — skip them and move on.
flush_interval.set_missed_tick_behavior(tokio::time::MissedTickBehavior::Skip);
loop {
tokio::select! {
biased;
result = transport.recv_media() => {
let pkt = match result {
Ok(Some(pkt)) => pkt,
Ok(None) => {
info!(%addr, participant = participant_id, "disconnected");
break;
}
Err(e) => {
error!(%addr, participant = participant_id, "recv error: {e}");
break;
}
};
if let Some(ref report) = pkt.quality_report {
metrics.update_session_quality(session_id, report);
}
let others = {
let mgr = room_mgr.lock().await;
mgr.others(&room_name, participant_id)
};
let pkt_bytes = pkt.payload.len() as u64;
for other in &others {
let peer_addr = other.connection().remote_address();
let fwd = forwarders
.entry(peer_addr)
.or_insert_with(|| TrunkedForwarder::new(other.clone(), sid_bytes));
if let Err(e) = fwd.send(&pkt).await {
let _ = e;
}
}
let fan_out = others.len() as u64;
metrics.packets_forwarded.inc_by(fan_out);
metrics.bytes_forwarded.inc_by(pkt_bytes * fan_out);
packets_forwarded += 1;
if packets_forwarded % 500 == 0 {
let room_size = {
let mgr = room_mgr.lock().await;
mgr.room_size(&room_name)
};
info!(
room = %room_name,
participant = participant_id,
forwarded = packets_forwarded,
room_size,
"participant stats (trunked)"
);
}
}
_ = flush_interval.tick() => {
for fwd in forwarders.values_mut() {
if let Err(e) = fwd.flush().await {
let _ = e;
}
}
}
}
}
// Final flush — send any remaining buffered packets.
for fwd in forwarders.values_mut() {
let _ = fwd.flush().await;
}
let mut mgr = room_mgr.lock().await;
mgr.leave(&room_name, participant_id);
}
/// Parse up to the first 2 bytes of a hex session-id string into `[u8; 2]`.
fn parse_session_id_bytes(session_id: &str) -> [u8; 2] {
let bytes: Vec<u8> = (0..session_id.len())
.step_by(2)
.filter_map(|i| u8::from_str_radix(session_id.get(i..i + 2)?, 16).ok())
.collect();
let mut out = [0u8; 2];
for (i, b) in bytes.iter().take(2).enumerate() {
out[i] = *b;
}
out
}
#[cfg(test)]
mod tests {
use super::*;
@@ -277,4 +480,97 @@ mod tests {
assert!(mgr.is_authorized("room1", Some("bob")));
assert!(!mgr.is_authorized("room1", Some("eve")));
}
#[test]
fn parse_session_id_bytes_works() {
assert_eq!(parse_session_id_bytes("abcd"), [0xab, 0xcd]);
assert_eq!(parse_session_id_bytes("ff00"), [0xff, 0x00]);
assert_eq!(parse_session_id_bytes(""), [0x00, 0x00]);
// Longer hex strings: only first 2 bytes taken
assert_eq!(parse_session_id_bytes("aabbccdd"), [0xaa, 0xbb]);
}
/// Helper: create a minimal MediaPacket with the given payload bytes.
fn make_test_packet(payload: &[u8]) -> wzp_proto::MediaPacket {
wzp_proto::MediaPacket {
header: wzp_proto::packet::MediaHeader {
version: 0,
is_repair: false,
codec_id: wzp_proto::CodecId::Opus16k,
has_quality_report: false,
fec_ratio_encoded: 0,
seq: 1,
timestamp: 100,
fec_block: 0,
fec_symbol: 0,
reserved: 0,
csrc_count: 0,
},
payload: Bytes::from(payload.to_vec()),
quality_report: None,
}
}
/// Push 3 packets into a batcher (simulating TrunkedForwarder.send),
/// then flush and verify all 3 appear in a single TrunkFrame.
#[test]
fn trunked_forwarder_batches() {
let session_id: [u8; 2] = [0x00, 0x01];
let mut batcher = TrunkBatcher::new();
// Ensure max_entries is high enough that 3 packets don't auto-flush.
batcher.max_entries = 10;
batcher.max_bytes = 4096;
let pkts = [
make_test_packet(b"aaa"),
make_test_packet(b"bbb"),
make_test_packet(b"ccc"),
];
for pkt in &pkts {
let payload = pkt.to_bytes();
let flushed = batcher.push(session_id, payload);
// Should NOT auto-flush — we are below max_entries.
assert!(flushed.is_none(), "unexpected auto-flush");
}
// Explicit flush (simulates the 5 ms timer tick).
let frame = batcher.flush().expect("expected a frame with 3 entries");
assert_eq!(frame.len(), 3);
for entry in &frame.packets {
assert_eq!(entry.session_id, session_id);
}
}
/// Push exactly max_entries packets and verify the batcher auto-flushes
/// on the last push (simulating TrunkedForwarder.send triggering a send).
#[test]
fn trunked_forwarder_auto_flushes() {
let session_id: [u8; 2] = [0x00, 0x02];
let mut batcher = TrunkBatcher::new();
batcher.max_entries = 5;
batcher.max_bytes = 8192;
let pkt = make_test_packet(b"hello");
let mut auto_flushed: Option<wzp_proto::packet::TrunkFrame> = None;
for i in 0..5 {
let payload = pkt.to_bytes();
if let Some(frame) = batcher.push(session_id, payload) {
assert!(auto_flushed.is_none(), "should auto-flush exactly once");
auto_flushed = Some(frame);
// The auto-flush should happen on the 5th push (max_entries = 5).
assert_eq!(i, 4, "expected auto-flush on the last push");
}
}
let frame = auto_flushed.expect("batcher should have auto-flushed at max_entries");
assert_eq!(frame.len(), 5);
for entry in &frame.packets {
assert_eq!(entry.session_id, session_id);
}
// Batcher should now be empty — nothing to flush.
assert!(batcher.flush().is_none());
}
}