The same internal research note's second research point develops an ML-based detector for Geneva-style automated censorship-evasion traffic; simple flow-level features (flow size, max packet size, RST/SYN/FIN flag counts, forward init-window bytes, inter-arrival timing) achieve near-perfect (ROC-AUC ~1.00) classification of Geneva-generated evasion traffic against CICIDS2017 and MAWI backbone background traffic using decision trees, LightGBM, XGBoost and random forest, with abnormal flow size (~150 bytes vs. 1000-30000 bytes typical) identified as the single most discriminative feature.
流大小: 正常流量通常具有较大的流大小（约1000-30000字节），而Geneva流的平均大小约为150字节...能够检测几乎所有的由Geneva生成的规避攻击流量，实现流级别的检测。
Defense implications
- Automated evasion-strategy-search tools (Geneva and similar) that produce abnormally small flows or unusual TCP flag distributions relative to real application traffic are flagged as ML-detectable in this research; any technique derived from such a search tool should be checked for flow-level statistical normality (size, flag counts, timing) — not just packet-content unobservability — before being trusted against a well-resourced adversary.
Related findings
A 2020 MESA Lab monthly report describes building a sapp plugin that extracts packet-sequence features specifically "for DoH (DNS-over-HTTPS) service discovery," alongside a broader CSTNET DoH measurement-report effort and configuring a DNS-to-DoH gateway -- confirming sapp is used to fingerprint DoH traffic via statistical sequence features rather than plaintext DNS content.
A 2024 MESA thesis proposal targets detecting "cross-border apps that illegally transmit personal information" inside encrypted traffic without decryption, by discovering app-specific tracking identifiers as <domain,parameter,value,interval> tuples and clustering flow structural similarity (Euclidean distance in a feature space) to separate device identifiers from background noise via a two-layer recognition model.
A MESA Lab student research report catalogs the GFW's known Shadowsocks-detection methodology (passive detection via first-packet length/entropy; active probing triggered after as few as 13 legitimate client connections, typically within seconds of the first legitimate connection) alongside six published ML-based Shadowsocks traffic-classification techniques (a packet-size-image CNN at >98% accuracy, random-forest on flow/host/DNS-behavior features, PCA-Pearson feature selection), compiled as apparent background research for in-house detection work.
A performance bug at a domestic Xinjiang test site (25-70Gbps) traced packet-processing lock contention to a plugin named "tsg_vulpes" calling an ONNX Runtime model for real-time "encrypted voice recognition" on live traffic; disabling this ML classifier resolved packet loss, indicating it's an optional, performance-costly add-on.
A dedicated internal "运营商前端分析团队" (Carrier Front-end Analysis Team) produces weekly, carrier-specific traffic-analysis reports explicitly titled "新疆移动流量分析报告" (Xinjiang Mobile Traffic Analysis Report) and "新疆联通流量分析报告" (Xinjiang Unicom Traffic Analysis Report), with instances dated from at least July 2022 through March 2024. One instance (Xinjiang Mobile, 2023-05-15) reports total carrier bandwidth of 4.4Tbps (1Tbps access bandwidth), 27.34PB of weekly throughput, 123 distinct identified application-layer protocols, and a section (4.1, pages 17+) explicitly tracking server IP/location for 28 named overseas apps (BBC, Discord, ESPN, Facebook, Gmail, Google, Hulu, Instagram, Netflix, Reddit, Skype, Telegram, Twitter, Uber, WhatsApp, Wikipedia, YouTube, Zoom, etc.).
MESA Lab researcher notes on an encrypted-video-identification project describe adding SSL-layer information output to more precisely trace a flow's true source/identity, and considering reinforcement learning so the identification model adapts as network conditions change, working within/around sapp's plugin limitations.