Optimizing Real-Time Audio/Video Transmission for 5G Remote Control Scenarios

5G remote‑control applications impose extremely strict requirements on real‑time audio‑video (AV) transmission, demanding end‑to‑end latency below 200 ms (ideally around 100 ms), jitter below 0.2 % and the ability to tolerate 20‑30 % packet loss. These metrics are far higher than those of typical video conferencing, live streaming, or surveillance.

The AV pipeline consists of capture, encoding, sending, transport, receiving, decoding and rendering. In this chain, the receiving side’s jitter buffer is the primary component responsible for both smoothing network fluctuations and contributing to overall latency.

Because modern chips have reduced the processing delay of encoding, decoding and rendering to under 10 ms, the dominant latency factors are the capture and transport modules, which are governed by camera hardware and network conditions. Therefore, joint optimization of the sending and receiving modules is essential.

The sending module typically uses RTP over UDP, incorporates congestion control, packet pacing and forward error correction (FEC). The receiving module contains a jitter buffer (handling out‑of‑order packets, frame detection and caching) as well as packet de‑multiplexing, error decoding and link‑state feedback for congestion control.

Key optimization goals are to shrink the jitter buffer size, which directly reduces latency. This requires minimizing packet loss and retransmission, especially the bursty loss caused by network congestion.

Common congestion‑control algorithms for real‑time AV include BBR and GCC. BBR estimates the bandwidth‑delay product but reacts slowly to sudden bandwidth drops, while GCC combines delay‑based and loss‑based signals and uses a Kalman filter to smooth delay estimates. For 5G air‑interface networks, which exhibit rapid bandwidth fluctuations due to varying signal‑to‑noise ratio (SNR) and scheduling cycles, a hybrid approach that augments GCC with SNR‑based estimates is recommended.

FEC techniques such as XOR coding or Reed‑Solomon (RS) codes add redundancy to recover random packet loss. However, short block lengths are insufficient for bursty loss typical of 5G uplink under deep fading. Introducing interleaving and adapting code length based on real‑time SNR can improve resilience without excessive bandwidth overhead.

Overall, by tailoring congestion control to 5G’s unique characteristics and optimizing the jitter buffer and FEC strategies, the end‑to‑end latency can be brought closer to the stringent 100 ms target, though further work is needed for cross‑region remote‑control scenarios.