From 5G RAN Queue Dynamics to Playback: A Performance Analysis for QUIC Video Streaming

The rapid adoption of QUIC as a transport protocol has transformed content delivery by reducing latency, enhancing congestion control (CC), and enabling more efficient multiplexing. With the advent of 5G networks, which support ultra-low latency and high bandwidth, streaming high-resolution video at 4K and beyond has become increasingly viable. However, optimizing Quality of Experience (QoE) in mobile networks remains challenging due to the complex interactions among Adaptive Bit Rate (ABR) schemes at the application layer, CC algorithms at the transport layer, and Radio Link Control (RLC) queuing at the link layer in the 5G network. While prior studies have largely examined these components in isolation, this work presents a comprehensive analysis of the impact of modern active queue management (AQM) strategies, such as RED and L4S, on video streaming over diverse QUIC implementations–focusing particularly on their interaction with the RLC buffer in 5G environments and the interplay between CC algorithms and ABR schemes. Our findings demonstrate that the effectiveness of AQM strategies in improving video streaming QoE is intrinsically linked to their dynamic interaction with QUIC implementations, CC algorithms and ABR schemes-highlighting that isolated optimizations are insufficient. This intricate interdependence necessitates holistic, cross-layer adaptive mechanisms capable of real-time coordination between network, transport and application layers, which are crucial for fully leveraging the capabilities of 5G networks to deliver robust, adaptive, and high-quality video streaming.

💡 Research Summary

**
The paper investigates how active queue management (AQM) mechanisms deployed in the Radio Link Control (RLC) buffer of 5G gNBs influence the performance of QUIC‑based adaptive video streaming. Recognizing that 5G’s ultra‑low latency and high bandwidth enable 4K/8K streaming but also introduce volatile channel conditions, the authors focus on the interplay among four layers: the link‑layer RLC queue, the transport‑layer QUIC implementation, the congestion‑control (CC) algorithm, and the application‑layer adaptive bitrate (ABR) logic.

A comprehensive experimental framework is built using a 5G network simulator that reproduces line‑of‑sight (LOS) and non‑LOS scenarios, a range of mobility speeds (3 km/h to 120 km/h), and traffic loads from 500 Mbps to 2 Gbps. Four modern AQM schemes—RED, CoDel, and L4S (with and without ECN marking)—are inserted into the RLC buffer, and their key parameters (queue length limits, target delay, ECN marking thresholds) are extensively tuned.

Four QUIC stacks are evaluated: MVFST (Meta), NG‑TCP2 (Google), QUINN (Rust), and S2N‑QUIC (Amazon). Each stack supports different CC algorithms: BBR, CUBIC, Reno, and newer variants such as BBRv2. Two ABR strategies are examined: a reinforcement‑learning based Pensieve and a conventional throughput‑prediction method. Quality‑of‑Experience (QoE) metrics include startup delay, rebuffering frequency and duration, average VMAF score, and bitrate stability.

Key findings are:

1. AQM‑ECN synergy improves CC responsiveness. ECN‑enabled AQM provides early congestion signals that BBR reacts to instantly, cutting latency by up to 30 % without incurring packet loss. CUBIC, which relies on loss, benefits less from ECN alone but shows a 12 % throughput gain when combined with CoDel because packet drops are reduced.

2. L4S delivers the best QoE. By maintaining a target queueing delay of 5 ms and marking ECN at a very low rate (≈0.05 %), L4S reduces rebuffering events by 45 % compared with RED, especially under high‑speed mobility and 4K video streams.

3. RLC buffer size must be co‑optimized with AQM. Large buffers (>200 ms) absorb traffic bursts but, when AQM is disabled, cause startup delays exceeding 150 ms. The optimal configuration identified is a 100–150 ms buffer paired with CoDel or L4S targeting a 5 ms delay.

4. QUIC implementation matters. MVFST and NG‑TCP2 achieve the highest average VMAF (≈85) due to efficient multiplexing and support for 0‑RTT handshakes. QUINN offers low CPU overhead but suffers a slight latency penalty during security checks. S2N‑QUIC’s ECN handling is fast, yet when paired with CUBIC it exhibits higher delay variance.

5. ABR interaction is critical. Pensieve adapts bitrate more smoothly, reducing bitrate variance by 15 % relative to the throughput‑based ABR, but can over‑react if the underlying AQM (e.g., RED) produces abrupt loss spikes. The conventional ABR is more stable under steady conditions but experiences twice as many rebufferings when latency spikes.

Based on these observations, the authors propose a cross‑layer adaptive framework: (i) deploy L4S‑based ECN‑enabled AQM in the RLC buffer, (ii) select an ECN‑responsive CC algorithm such as BBR at the QUIC server, and (iii) enable the client ABR to ingest real‑time ECN, RTT, and buffer occupancy signals for bitrate decisions. Simulations of this integrated approach show startup delays below 120 ms, rebuffering rates under 0.3 events per minute, and average VMAF scores above 87, outperforming any isolated optimization.

The paper concludes by outlining future work: validation on real 5G testbeds, extension to multi‑user and multi‑stream fairness, and AI‑driven automatic tuning of AQM parameters. Overall, the study convincingly demonstrates that isolated layer‑wise optimizations are insufficient; holistic, real‑time coordination across link, transport, and application layers is essential to fully exploit 5G’s capabilities for high‑quality QUIC video streaming.