experiments

Experiments

We conduct a small initial measurement study. The results of which are categorized below based on the two types of anchors used.

results/local.csv contains domain names that belong to educational institutions. Filtered if they have .edu as their tld or .ac as their sld
results/local-filtered.csv contains the sublist from the first 2852 domains in local.csv that return correct time
results/anchors.csv contains those that are not backed by a CDN provider -> 298 domains remain

Anchor owner	Name	IP (Location)	RTT AWS-CA	RTT AWS-NV	RTT AWS-SP
Google	roughtime.sandbox.google.com	173.194.206.158 (Virginia)	26-27ms	10-11ms	155ms
Cloudflare	roughtime.cloudflare.com	162.159.200.1 (SF)	30-33ms	17-18ms	30ms
Chainpoint	roughtime.chainpoint.org	35.194.78.198 (Washington DC)	?	3.4-3.5ms	208ms
int08h	roughtime.int08h.com	35.192.98.51 (Iowa)	?	30ms	190ms

time_rsa_filter.sh to find domains that return correct time and support RSA authentication
iploc_keycdn.py to find ISP information (uses ip2location's DB) and iploc_ip2location.py to fetch IP location information
rtt_openssl.sh and rtt_tlse.sh to find round trip times

Data inside results is organized as follows:

exp1: Finding anchors close to three AWS locations: London, North Virginia(NV), Singapore
exp2: Observing public keys used by TLS servers
exp3: Measuring TLS RTT from AWS-NV to close anchors over a period of time
exp4: Running GeoCommit from AWS-NV and AWS-Lon for over a week with 10 selected anchors