quic protocol

This study systematically investigates the Quality-of-Experience (QoE) impact mechanisms of the QUIC protocol in multi-client video streaming scenarios. Focusing on two representative use cases—video-on-demand (VoD) and low-latency live (LLL) streaming—we employ a trace-driven simulation framework to analyze cross-layer interactions between mainstream QUIC implementations (featuring congestion control algorithms including Cubic and BBR) and adaptive bitrate (ABR) strategies (BOLA, Pensieve). We empirically reveal, for the first time, that identical congestion control algorithms exhibit substantial performance variation across different QUIC implementations, leading to measurable discrepancies in key QoE metrics—namely, stall ratio, startup latency, and average bitrate. Building upon this insight, we propose a QUIC–ABR cross-layer co-optimization framework that jointly adapts congestion window feedback and bitrate selection decisions. Our approach achieves quantifiable QoE improvements: a 32% reduction in stall ratio and an 18% increase in average video quality.

Optimizing Quality of Experience (QoE) for 4K+ video streaming over 5G networks remains challenging due to complex cross-layer interactions among application-layer Adaptive Bitrate (ABR) algorithms, transport-layer QUIC congestion control, and link-layer RLC queue dynamics. Method: This paper proposes a full-stack co-adaptive mechanism that jointly models RLC buffer dynamics and integrates multi-QUIC framework simulations to quantify the interplay among Active Queue Management (AQM) policies (e.g., RED/L4S), QUIC implementations, congestion control algorithms, and ABR schemes. Contribution/Results: It is the first work to systematically characterize the deep coupling between AQM strategies and both QUIC stack behavior and ABR decision logic. Experimental results demonstrate that isolated layer-wise optimization yields marginal gains, whereas cross-layer coordination significantly enhances streaming stability and QoE—enabling more robust adaptive video delivery under high-bandwidth, low-latency 5G conditions.

This study addresses the vulnerability of MoQ’s default bitrate switching mechanism to playback interruptions under network congestion, which particularly degrades the time-shifted streaming experience. The authors evaluate SWITCH-style adaptive bitrate (ABR) algorithms in a Mininet emulation environment, demonstrating for the first time that standard ABR algorithms—without modification—can effectively support MoQ time-shifted playback. Experimental results reveal a counterintuitive yet significant increase in throughput following rebuffering events, suggesting that existing ABR strategies can directly enhance quality of service for time-shifted scenarios. These findings not only validate the immediate applicability of conventional ABR logic to MoQ-based time-shifted streaming but also highlight key directions for optimizing ABR switching behavior within the MoQ framework.

This paper addresses pacing inaccuracy in QUIC user-space implementations, caused by coarse-grained timers, system call overhead, and scheduling latency. We systematically evaluate pacing performance across three major implementations—quiche, picoquic, and ngtcp2—and quantify the impact of Linux qdiscs (especially FQ), GSO hardware offloading, and the Earliest Transmit First (ETF) scheduler. We present the first empirical comparison of pacing accuracy between pure user-space and kernel-assisted approaches. Furthermore, we propose and validate a kernel patch that leverages per-packet timestamping within the GSO buffer to enable fine-grained scheduling while preserving batch-processing efficiency. Experiments show that picoquic with BBR achieves sub-millisecond pacing accuracy in pure user-space mode; the FQ qdisc provides the most favorable pacing environment for QUIC; and kernel-enhanced GSO pacing reduces timing jitter by 40%, significantly improving real-time audio/video quality.

This paper addresses the challenge that conventional TCP applications cannot leverage QUIC’s advantages—such as built-in encryption, stream multiplexing, and rapid loss recovery. To bridge this gap, we propose and implement a lightweight TCP-over-QUIC streaming tunnel, built in userspace using Rust and the Quinn QUIC library. It transparently proxies TCP connections and maps them onto QUIC streams without requiring modifications to legacy TCP applications. To our knowledge, this is the first systematic performance evaluation of such a tunnel under realistic network impairments. Experiments show that under 20% packet loss, the tunnel achieves significantly higher throughput than native TCP and exhibits superior stability under high latency and packet reordering. In ideal networks, it incurs only acceptable overhead from encryption and stream scheduling. Our core contribution is the empirical validation of QUIC-based TCP tunneling as both feasible and practically effective, delivering a low-overhead, robust engineering implementation.

This work addresses the challenge of starvation faced by high-priority small flows in mixed-criticality applications sharing a single TCP connection, where parallel connections are often constrained by network policies and fail to ensure per-flow fairness. The authors propose CATS—a Conductor-driven asymmetric transport scheme—that operates atop TCP to simultaneously guarantee non-starvation for critical flows, per-flow fairness, and bounded inter-class latency within a single connection. By introducing sender-side credit shaping, dynamic priority tagging, and sequence number reordering, CATS overcomes inherent limitations of in-band prioritization. The study further identifies three fundamental structural constraints of in-band priority mechanisms and presents a systematic solution leveraging QUIC’s multi-stream architecture to achieve intra-endpoint flow isolation and aggregated congestion control. ns-3 simulations and a QUIC-based prototype demonstrate that CATS effectively preserves critical traffic latency, while CATS-over-QUIC transcends structural limitations to deliver strict QoS guarantees under per-flow fairness.

This study addresses the degradation of Quality of Experience (QoE) in MPEG-DASH video streaming over integrated 5G Integrated Access and Backhaul (IAB) networks that incorporate Low Earth Orbit (LEO) satellites, where high latency and throughput fluctuations pose significant challenges. The authors develop an end-to-end simulation framework based on ns-3 to systematically evaluate, for the first time, the impact of TCP and QUIC protocols—each paired with various congestion control algorithms including CUBIC, NewReno, and BBR—on video QoE. Experimental results demonstrate that the QUIC-BBR combination achieves superior performance in terms of playback duration, rebuffering frequency, and latency, significantly outperforming other protocol-algorithm pairings. This configuration strikes an optimal balance among throughput, delay, playback continuity, and fairness, offering an effective optimization pathway for video delivery in satellite-integrated internet architectures.

As per-port bandwidth in data center switches continues to increase, traditional buffer-sharing strategies suffer from degraded performance and high complexity. This work proposes BShare, a lightweight, queueing-delay-based buffer-sharing mechanism that integrates buffer management with active queue management using only a single configurable parameter. By dynamically monitoring queueing delay to adjust buffer allocation in real time, BShare significantly simplifies policy design while maintaining compatibility with advanced transport protocols such as PowerTCP. Simulation results demonstrate that under burst-intensive workloads, BShare reduces flow completion time (FCT) by up to 45.07% compared to the ABM scheme.

Existing transport-layer hardware struggles to flexibly support the evolution of new protocols due to rigid protocol logic or reliance on protocol-specific assumptions. This work proposes PITA, a novel architecture that reconfigures core components—such as scheduling, packet generation, and data reassembly—through a unified event–state–instruction abstraction model, enabling a protocol-agnostic and line-rate programmable transport-layer datapath. By eliminating protocol-specific assumptions, PITA efficiently supports semantically diverse protocols, including TCP and RoCE, on a single FPGA (Alveo U250) while fully preserving their end-to-end behavioral differences. Experimental results demonstrate that the system meets timing constraints at 250 MHz with low hardware overhead and excellent performance.