A KL-divergence classifier trained to distinguish CovertCast streams from real YouTube streams achieved only 33–45% true positive rate on packet-size distributions and 36–41% on inter-packet timing distributions — below random guessing — while maintaining 86–98% true negative rates. Overall classifier accuracy was approximately 65–68%, driven entirely by the high true negative rate rather than genuine detection capability.

CovertCast's broadcast model decouples server workload from client count: one server can serve unlimited simultaneous clients without per-connection overhead, unlike hide-within systems such as FreeWave where server costs grow linearly with users. This architecture also defeats Sybil-based DoS attacks, because flooding the server with fake client requests does not increase server load — the server never processes individual client connections.

§2, §7.2 defense

Under degraded network conditions, CovertCast page load times increased by 2–3× at 800 Kbps (below YouTube's minimum 720p bitrate of 1.5 Mbps), with 20 of over 4,000 images dropped at 800 Kbps; at 10% packet loss, 35 images were missed due to YouTube temporarily accelerating video playback; at 20% packet loss, 720p video could not be loaded at all.

§6.4, Fig. 6, Fig. 7 evaluation

CovertCast uses the identical video codecs, streaming protocols (RTMP/HTTPS), and server endpoints as any other YouTube live stream, making it indistinguishable from regular streaming traffic to both passive protocol-analysis and active traffic-manipulation attacks. Any active attack that disrupts CovertCast connections — such as selective packet dropping — would equally disrupt all non-circumvention viewers of the same streaming service, imposing prohibitive collateral damage.

§7.4, §7.5, §7.6 defense

Because CovertCast clients connect to live-streaming service infrastructure (e.g., YouTube servers) rather than to CovertCast servers directly, IP-address blacklisting of CovertCast infrastructure does not allow censors to identify or disrupt client connections. Discovering the CovertCast server's IP address is therefore irrelevant to the censor's blocking goal.

§7.3 defense

The host-profiling censor (passive traffic analysis: count connections per server, block those exceeding a threshold τ within a window w) blocks essentially all circumvention user traffic within 30 time steps for all classifier qualities tested (ρ_TP ∈ {0.9, 0.95, 0.99}), while causing far less collateral damage than zig-zag (never exceeding ~30% innocent server blocking). Snowflake resists this attack well: with w=3, τ=3, over 94.48% of users receive a proxy within 2 steps even with worst-classifier rules, and final unblocked server rates are 91.24–99.04%. The host profiling approach is feasible for passive censors who cannot enumerate the distribution channel.

2026-fares-game §5.2, Table 2, Table 3 traffic-shapeml-classifier

Balboa's synchronous leaf-content replacement adds non-negligible timing differences that allow censors to identify its activity with up to ~90% accuracy over different network conditions. The timing anomaly arises because Balboa performs data substitution directly at each data exchange, delaying the server's response while covert data is prepared.

2026-kamali-huma §I (Introduction), §II-A traffic-shapeml-classifier

Huma's deferred-reply / double-request receive (DRR) protocol reduces a traffic-fingerprinting XGBoost classifier's accuracy to at most 54% (near random guessing) across geographically distributed clients (San Francisco, Frankfurt, Bangalore). A Kolmogorov-Smirnov test on absolute page-load timing distributions yields D=0.03, p=0.98 for U.S. clients — substantially tighter than Waterfall of Liberty's D=0.11 at p=0.5 — confirming that Huma flows are statistically indistinguishable from benign HTTPS fetches.

2026-kamali-huma §V-D, Table II traffic-shapeml-classifier

The PPBR (probabilistic profile-based routing) protocol leaks user community membership through observable routing decisions: in a controlled experiment with 800 majority and 200 minority users, a statistical disclosure attack achieved a true positive rate of 100% and false positive rate of 0% when identifying minority users. Even under a conservative PPBR configuration (top 1/3 fraction acceptance), the attack achieved 100% TPR and only 0.4% FPR.

2026-ratliff-mirage §V ml-classifiertraffic-shape

Banking apps from major Russian institutions (Sber, T-Bank, VTB, Alfa-Bank) combine VPN detection with behavioral biometrics — screen pressure, touch coordinates, and gesture timing — enabling cross-account re-identification of users behind proxies. 11 apps received a "RED" (maximum surveillance) rating. T-Bank, Yandex services, and MAX additionally deploy active anti-analysis features that detect research tooling on the device (rooted devices, emulators, Frida, etc.).

2026-rks-russian-apps-vpn-detection §5, §6 ml-classifiertraffic-shape

Amnesty International's 102-page investigation identifies a multi-vendor surveillance stack deployed in Pakistan: Chinese DPI (Geedge/MESA-derived), Canadian social-media monitoring (Netsweeper), and Emirati commercial spyware (Pegasus and FinFisher). The system enables deep packet inspection, SNI-based filtering, and traffic-shape classification at national scale, including targeted interception of encrypted messaging apps and VPN traffic.

2025-amnesty-pakistan-shadows §3, §5 dpisni-blockingip-blockingtraffic-shapeml-classifier