An adversary aware of a decoy router's location can force decoy-routed flows to be unprocessable by fragmenting all packets below the size of a complete TCP header in the first fragment, preventing flow assignment and forcing the router into expensive reassembly. Alternatively, the adversary can use small-fragment attacks to grow the router's state table, analogous to NAT resource exhaustion. The paper identifies fragmentation-based denial as a harder-to-mitigate attack class than sentinel replay.

Decoy routing places the circumvention service at transit routers rather than fixed-IP edge proxies, so the client addresses packets to any reachable decoy destination and the router hijacks the flow on the client's behalf. A single well-placed router may lie on paths to millions of destinations, making circumvention proxies appear ubiquitously deployed from an adversary's perspective. Blocking such a router requires disrupting ordinary traffic for large fractions of the Internet, qualitatively raising the cost of IP-address-based censorship.

§1.2 defense

A preplay attack defeats the TLS-sentinel covert channel: the adversary intercepts each ClientHello, immediately sends a copy to the decoy destination before the client's copy arrives, causing the sentinel to be consumed and poisoned. The client can never establish a decoy routing session while ordinary TLS to the decoy destination continues to work normally, giving the adversary both blocking capability and forensic confirmation that decoy routing was attempted. The paper notes this vulnerability is specific to the TLS sentinel and that alternatives such as port-knocking sentinels may not share it.

§4.1 detection

TCP flow hijacking by the decoy proxy is practical under an asymmetric routing assumption: expected sequence numbers are recoverable from ACK values in client-originated packets alone, so the decoy router need not observe return traffic. The proxy forges a TCP RST to the decoy destination and mimics its TCP options (timestamp, window scale, SACK) to reduce detectability; these options are conveyed encrypted inside the sentinel's 28-byte TLS random field.

§3.3 defense

Clients embed HMAC-derived, time-varying sentinels into the 28-byte random field of the TLS ClientHello message, which decoy routers can scan at line rate. Sentinels are keyed to the current hour and a per-hour sequence number, providing freshness. This covert channel requires no out-of-band signaling and is invisible to passive observers who see only a normal TLS handshake toward the decoy destination.

§3.2 defense

Dominant failure modes differ systematically by country: China (AS45090) shows connect timeouts in 75% of DoT failures (IP-level blocking); Kazakhstan (AS48716) shows post-TLS-handshake timeouts in 72% of DoT failures (likely ACK or segment discard after handshake); Iran (AS197207) shows TLS handshake timeouts in 80% of DoT failures. Packet capture analysis confirmed that timeouts during and after the TLS handshake correspond to unacknowledged TCP segments, not connection resets.

2021-basso-measuring §V-F, §V-G, Table VIII ip-blockingtls-fingerprintmiddlebox-interference

Residual censorship—where a censor continues blocking all traffic on a 3- or 4-tuple after an initial censorship event—is active in China (HTTP: 90s 3-tuple RST injection; ESNI: 120–180s 3+4-tuple null routing), Iran (HTTP+SNI: 180s 4-tuple null routing, occasionally up to 5 minutes; protocol filter: 60s), and Kazakhstan (HTTP+SNI: 120s 4-tuple null routing). A December 2020 Quack scan found 3-tuple stateful disruption in 33 countries and null-routing censorship in 18, suggesting much broader applicability.

2021-bock-your §IV, Table I rst-injectionip-blockingmiddlebox-interference

All tested censors (China, Iran, Kazakhstan) can be triggered statelessly—without completing a TCP 3-way handshake—using a SYN with decremented sequence number followed by a PSH+ACK containing the forbidden payload. This stateless triggering enables fully off-path, source-spoofed attacks: an adversary with packet-spoofing capability can residually censor a victim pair they have no on-path access to.

2021-bock-your §IV, §V.A rst-injectionip-blockingmiddlebox-interference

Iran and Kazakhstan reset the residual censorship timer whenever the censor observes any matching packet from the victim, so TCP retransmissions from the victim's own stack inadvertently extend the blocking window far beyond the nominal 120–180s. China's HTTP residual censorship has only ~50% per-request reliability from some vantage points due to heterogeneous GFW middlebox load-balancing, but reliability plateaus near 100% after 7 repeated censorship triggers sent ahead of time.

2021-bock-your §IV (timer reset), §IV (reliability), Fig. 2 ip-blockingmiddlebox-interferencerst-injection

All three prior end-to-middle (E2M) schemes — Telex, Cirripede, and Decoy Routing — require an inline flow-blocking component at the participating ISP, which adds latency, introduces a single point of failure, and may violate carrier SLAs. In private discussions with ISPs, the authors found that despite willingness to assist Internet freedom technically and financially, none were willing to deploy existing E2M technologies due to these operational impacts. TapDance removes the inline blocking requirement entirely, requiring only a passive tap and packet-injection capability.

2014-wustrow-tapdance §1, §6 middlebox-interferenceip-blocking

A hybrid two-stage blocking system (IP-redirect first stage, URL-proxy second stage) can be exploited as an oracle to enumerate blocked IP addresses by sending TCP packets with a TTL sufficient to reach the first-stage redirector but insufficient to reach the destination. Non-redirected IPs return ICMP TTL-expired from an intermediate router, while redirected IPs return a SYN/ACK from the web proxy impersonating the destination. A live scan of a /24 subnet confirmed 17 redirected IP addresses, yielding 91 associated hostnames across 9 of those IPs.

2006-clayton-failures §5.2 ip-blockingmiddlebox-interference