latency · jitter · packet loss to the same origin form a stable signature for your isp + path · runs locally
how it works
every network path has a unique timing signature — shaped by your router, the isp’s peering, the cdn edge, and ambient congestion. by sending many small same-origin requests and measuring RTT with sub-millisecond precision we can build a statistical fingerprint: mean latency, jitter, p95 spike height. the pattern is stable enough to distinguish home wifi from coffee shop 4g.
probe
why this matters
- network path fingerprinting has been used to distinguish devices behind NAT — each device has slightly different buffering, congestion response, and ack timing, even on the same ip.
- cdn attribution: the p95 spike height is often diagnostic of which cdn edge you hit — cloudflare vs akamai vs fastly each have different topologies.
- corporate vs consumer: traffic-shaped enterprise networks show dramatically different jitter profiles compared to unconstrained home broadband.