Online Advertising under Internet Censorship

Ad server domains are structurally immune to censor blocking due to collateral-damage risk: Google DoubleClick is embedded in 1,843,854 publisher sites and PubMatic in 215,046, making IP-blocking of these domains prohibitively costly for any censor. Measurements of Alexa top-10K confirm the top 20 ad servers handle more than 75.6% of all ad requests.

§4.2, §4.3 deployment ip-blocking generic

82.2% of ad requests from Alexa top-500 websites are sent over HTTPS (Table 2), encrypting the HTTP Referer field. This prevents censors from correlating a user's direct-path ad request back to a censored publisher domain in the vast majority of cases; only the remaining 17.8% of HTTP ad requests are vulnerable to Referer-based traffic analysis.

§4.3, Table 2 evaluation dpi generic

Relay-based circumvention severely degrades ad relevance: across Alexa top-500 uncensored sites, the overlap between ad sets fetched via Tor and the direct-path ground truth averaged only 28%, with near-zero overlap for sites serving geo-targeted ads. For blocked sites, only ~16% of ads shown via Tor were in the user's language.

§3.2, Table 3 evaluation ip-blockingdpi generic

ADVENTION's split-path design — fetching publisher content via relay and ad requests via the direct path — raises average ad-set overlap from 28% (Tor) to 70%; combining ADVENTION with Intelligent Relay Selection (language-matched relay) further increases average overlap to ~80%. For blocked sites, ADVENTION with IRS raised ad relevance from ~16% to 100%.

§5.1, §5.3, §5.4 evaluation ip-blocking generic

ADVENTION provides up to 47% improvement in average page load time (PLT) compared to Tor, because ad requests — which are often on the critical rendering path — are served over the direct channel rather than through the relay. The exact improvement depends on webpage structure and bottleneck resources.

§5.1, Figure 4 evaluation generic