FlowPaint evades all three GFW-style rule-based middlebox configurations — stateful TCP keyword reassembler (Censor A), heuristic header filter dropping flows with TCP data offset ≥ 9 (Censor B), and anti-replay validator caching (SEQ, payload_hash) pairs and injecting bidirectional RSTs on replay detection (Censor C). Geneva's duplicate-packet desynchronization strategy fails against Censor C because its malformed artifacts simultaneously satisfy the anti-replay blocking condition, while FlowPaint succeeds by recalculating checksums and injecting no extra packets.

Under maximum-sensitivity classifier settings, FlowPaint forces Random Forest to accept ~25% FPR across all four out-of-distribution protocols in order to maintain high TPR — blocking 1 in 4 legitimate flows, a level of collateral damage operationally unsustainable for state-level censors. Obfs4 and UPGen exhibit clean separation (TPR ≈ 1.00, FPR ≈ 0.00), imposing no such dilemma on the censor and thus offering no protection from the censor's perspective either.

§5.1.2, Table 2 evaluation

FlowPaint achieves a steady-state goodput of 3.05 Gbps versus 10.7–10.85 Gbps for lightweight baselines (UPGen, Dummy) that provide near-zero evasion (TPR ≈ 1.00), with an initial TTFB of ~1.9 s dominated by fixed diffusion model inference cost that grows by only 1.28× as network RTT increases tenfold (20 ms to 200 ms). At 3.05 Gbps, FlowPaint exceeds 4K and 8K streaming bandwidth requirements (25 Mbps and 100 Mbps respectively) by two orders of magnitude, indicating that compute overhead is a manageable hardware-scaling constraint rather than a fundamental barrier.

Appendix A, Table 7, Fig. 4 evaluation

Under a strict FPR ≤ 0.1% operational budget, FlowPaint achieves a TPR of 0.00 against Decision Tree and nPrintML classifiers across multiple out-of-distribution protocols (CurveZMQ, secio, SSH, TLS), while Obfs4 and UPGen remain near-fully detectable (TPR ≈ 1.00). Even against Deep Fingerprinting, FlowPaint reduces TPR to 0.80–0.92, compared to 1.00 for Obfs4 across all protocols tested.

§5.1, Table 1 defense

Obfs4's payload randomization creates a high-entropy statistical profile that a Decision Tree detects with TPR ≈ 1.00 and FPR ≈ 0.00 across all four out-of-distribution protocols tested. UPGen's procedural protocol assembly similarly yields TPR ≈ 1.00 for most classifiers, confirming that both entropy randomization and rigid structural assembly leave artifacts detectable by even simple statistical classifiers operating under strict false-positive constraints.

§5.1.3, Table 2 detection

The Reality protocol triggers immediate RST flood responses from Iranian ISP infrastructure at the completion of the TLS handshake, suggesting Iranian DPI has developed a fingerprint or active-probe response specifically targeting Reality's TLS handshake forwarding mechanism as of June 2026.

2026-pooribitwise-iran-advanced-dpi Issue body — Other Observations §2 dpirst-injectiontls-fingerprintactive-probing

By 2018 the GFW shifted from blocking Tor bridges by (IP, port) tuples to blocking the entire IP address. A blocked bridge remains inaccessible for exactly 12 hours; the block renews to 12 hours if any additional Tor connection attempt is made during that window, after which the GFW re-scans and removes the IP from the blacklist if Tor is no longer running.

2018-dunna-analyzing §4.2 dpiactive-probingip-blockingrst-injection

HTTP censorship blockpages in Iran fell into two distinct clusters distinguished by IP TTL values: type NTL1 (TTL 53–55) and type NTE1 (TTL 186–188); TCP RST injections targeting TLS and DNS-over-TCP exhibited the same two TTL ranges. The NTL1 injector was absent on June 11 but present on other days, leading the authors to conclude that an entire injector went offline rather than that a single injection pattern was selectively disabled.

2026-anon-insights-into-iranian §2.2.5 rst-injectionpacket-injectiondpi

The blog author, drawing on evaluation experience, concludes that LetsVPN's failure was not caused by IP exhaustion or ordinary node instability but by precise protocol-signature identification: once the GFW extracts a client handshake feature, it can simultaneously block all connections sharing that signature across hundreds of thousands of users.

2026-anon-letsvpn-vpn §公告里的几句大实话 dpitls-fingerprintactive-probing

As of June 2026, Iranian DPI blocking timescales are: CDN-fronted VLESS+WebSocket domains blocked within 24 hours; SSH dynamic port forwarding server IPs blocked within 8 hours; Reality and WireGuard connections blocked immediately (seconds) after handshake completion. This graduated timeline suggests separate fast-path (handshake-pattern) and slow-path (domain reputation / traffic accumulation) detection pipelines operating in parallel.

2026-pooribitwise-iran-advanced-dpi Issue body — combined observations dpisni-blockingip-blockingrst-injectiondns-poisoning

Re-testing in 2025 on a Pixel 10 Pro XL running Android 16 with October 2025 security updates confirmed that blind in/on-path VPN inference attacks remain fully viable despite CVE-2019-9461, CVE-2019-14899, and CVE-2024-49734 having been formally closed. All three core attack primitives—VPN-assigned internal IP discovery, active connection inference, and TCP reset injection via sequence/acknowledgment window scanning—succeeded across OpenVPN, WireGuard, and NordLynx.

2026-tolley-architectural §5.1–5.3 traffic-shaperst-injectionactive-probing