AV1 MPEG-TS Multiplexing Jitter Tolerances

This article provides an overview of the technical limits and timing specifications governing audio-video multiplexing jitter when transmitting AV1 elementary streams over MPEG Transport Stream (MPEG-TS). It covers the normative limits defined by the ISO/IEC 13818-1 Transport Stream System Target Decoder (T-STD) model, detailing allowable Program Clock Reference (PCR) jitter, packet arrival jitter, presentation timestamp (PTS) timing offsets, and the standard tolerances for end-to-end audio-to-video synchronization (lip-sync).

The Timing Framework: ISO/IEC 13818-1 T-STD

Carrying an AV1 stream inside an MPEG-TS container requires adherence to the standard ISO/IEC 13818-1 timing and buffer models (T-STD), alongside the specific carriage specifications defined for AV1 (such as those established by the Alliance for Open Media and modern digital television standards like DVB). Timing relies on a 27 MHz System Time Clock (STC) shared between the multiplexer and demultiplexer.

Because AV1 uses complex frame structures—including show-existing frames, temporal units, and large variance between intra-frames and inter-frames—strict multiplexing jitter boundaries are necessary to prevent decoder buffer overflow and underflow.

Program Clock Reference (PCR) Jitter Limits

The foundational timing mechanism in MPEG-TS is the PCR, which synchronizes the receiver's local STC oscillator.

• Maximum PCR Jitter: The normative standard limits PCR jitter to ±500 nanoseconds relative to an ideal linear 27 MHz time base at the output of the multiplexer.
• Repetition Rate: PCR values must be inserted at least once every 100 milliseconds, although broadcast and streaming applications typically require an interval of 40 milliseconds or less to maintain clock lock.
• Measurement Standard: As specified in ETSI TR 101 290 (First Priority metrics), PCR overall jitter (PCR_OJ) must not exceed ±500 ns to avoid phase-locked loop (PLL) tracking issues in downstream demultiplexers.

Packet Arrival Jitter and Buffer Tolerances

Network delivery and multiplexing operations inherently introduce packet delay variation (PDV).

• Multiplex Packet Jitter: In a CBR (Constant Bitrate) MPEG-TS delivery, packet arrival timing must stay within the compliance window of the smoothing buffer (\(SB_n\)) and elementary stream buffer (\(EB_n\)). The accepted tolerance for arrival deviation on a per-packet basis is typically bounded within ±1 millisecond in traditional transmission networks, with IP network buffers generally sized to absorb up to 50 ms to 100 ms of transit packet jitter before passing to the demux layer.
• Buffer Residence Time: The maximum allowable delay between the arrival of an AV1 packet at the input buffer and its presentation time (determined by PTS/DTS) must not exceed 1.0 second, except under specific extended profile definitions.

Audio-Video Synchronization (PTS/DTS Skew)

A/V multiplexing jitter directly impacts the temporal relationship between AV1 video and its associated audio tracks (such as AAC, AC-4, or Opus). The multiplexer must align Presentation Time Stamps (PTS) according to established international broadcasting limits:

• Source Multiplexer Alignment: At the point of multiplexing, the difference between audio PTS and video PTS for the corresponding acoustic/visual event must not deviate by more than ±5 milliseconds.
• Permissible Delivery Asymmetry: Following multiplexing and network transport, the operational tolerances defined by ITU-R BT.1359 and EBU R37 specify:
  • Audio leads Video (Advanced): Maximum allowable jitter/skew is +15 milliseconds.
  • Audio lags Video (Delayed): Maximum allowable jitter/skew is -45 milliseconds.

Any jitter causing the timing stamps to drift beyond these boundaries results in perceptible lip-sync errors or buffer underflows, causing playback artifacts and decoder pipeline stalls.

AV1-Specific Considerations

AV1 elementary streams carried over MPEG-TS utilize dedicated metadata descriptors and transport packet formatting to map Operating Points and Temporal Units into Packetized Elementary Stream (PES) packets.

Due to the bursty nature of AV1 compressed data compared to legacy codecs, multiplexers must apply rate-smoothing algorithms to leak AV1 transport packets into the transport multiplex at a stable leak rate (\(R_{vbi}\)). Failure to precisely schedule these packets causes transient buffer spikes, driving PCR and DTS jitter past the normative ±500 ns and ±5 ms bounds.