Undermining Delay-based QUIC Congestion Control: A Receiver-driven Attack via Crafted Host Delays
QUIC gains significant attention due to its superior transmission performance, achieving widespread adoption in both academia and industry. To improve round-trip time (RTT) estimation, QUIC introduces the Host Delay field, enabling senders to exclude receiver-induced delays. Many delay-based congestion control algorithms (CCAs) rely on these RTT estimates to detect congestion and regulate sending rates. However, we find that malicious Host Delay values can distort RTT measurements, causing inappropriate rate adjustments by CCAs.In this paper, we investigate a new class of attacks leveraging maliciously crafted Host Delay values. To our knowledge, we are the first to analyze the vulnerability of Host Delay and present QUDIT, a universal receiver-driven attack targeting delay-based QUIC CCAs. Unlike prior attacks that presume full network queuing delays visibility and undetected injection capabilities, our attacker model only grants the adversary access as a standard QUIC receiver with limited knowledge of bottleneck conditions. We minimally modify the QUIC receiver to infer bottleneck queuing behavior in real time. Based on these inferences, we design dynamic Host Delay crafting strategies tailored to the specific behavior of various delay-based CCAs and accounting for random network fluctuations. Our attack prompts the sender to overshoot its rate, leading to excessive bandwidth consumption at the bottleneck and degradation of competing flows. Results demonstrate the throughput degradation of victim flow achieves up to 60% within 0.3 s and amplification gains between 200× and 600×. We propose defenses mitigating QUDIT, with vulnerabilities reported to IETF and QUIC maintainers.
Authors
- Jianping Wu (ORCID: https://orcid.org/0000-0002-8887-8031)
- Enhuan Dong (ORCID: https://orcid.org/0000-0002-2539-8241)
- Jia Zhang (ORCID: https://orcid.org/0000-0002-2513-4115)
- Jiahao Cao (ORCID: https://orcid.org/0000-0001-7139-376X)
- Shaorui Ren (ORCID: https://orcid.org/0009-0007-6782-6156)
- Yixuan Zhang (ORCID: https://orcid.org/0000-0001-5440-296X)
- Mingwei Xu (ORCID: https://orcid.org/0000-0002-4847-4585)
Institutions
- Tsinghua University (CN)
Publication Details
- Published
- 2025-09-22
- DOI
- https://doi.org/10.1109/icnp65844.2025.11192451
- Primary Topic
- Software-Defined Networks and 5G
- Type
- article
- Field-Weighted Citation Impact
- 0.00